add idl4k kernel firmware version 1.13.0.105

kernel/drivers/infiniband/ulp/ipoib/Kconfig

@@ -0,0 +1,50 @@
+config INFINIBAND_IPOIB
+	tristate "IP-over-InfiniBand"
+	depends on NETDEVICES && INET && (IPV6 || IPV6=n)
+	select INET_LRO
+	---help---
+	  Support for the IP-over-InfiniBand protocol (IPoIB). This
+	  transports IP packets over InfiniBand so you can use your IB
+	  device as a fancy NIC.
+
+	  See Documentation/infiniband/ipoib.txt for more information
+
+config INFINIBAND_IPOIB_CM
+	bool "IP-over-InfiniBand Connected Mode support"
+	depends on INFINIBAND_IPOIB
+	default n
+	---help---
+	  This option enables support for IPoIB connected mode.  After
+	  enabling this option, you need to switch to connected mode
+	  through /sys/class/net/ibXXX/mode to actually create
+	  connections, and then increase the interface MTU with
+	  e.g. ifconfig ib0 mtu 65520.
+
+	  WARNING: Enabling connected mode will trigger some packet
+	  drops for multicast and UD mode traffic from this interface,
+	  unless you limit mtu for these destinations to 2044.
+
+config INFINIBAND_IPOIB_DEBUG
+	bool "IP-over-InfiniBand debugging" if EMBEDDED
+	depends on INFINIBAND_IPOIB
+	default y
+	---help---
+	  This option causes debugging code to be compiled into the
+	  IPoIB driver.  The output can be turned on via the
+	  debug_level and mcast_debug_level module parameters (which
+	  can also be set after the driver is loaded through sysfs).
+
+	  This option also creates a directory tree under ipoib/ in
+	  debugfs, which contains files that expose debugging
+	  information about IB multicast groups used by the IPoIB
+	  driver.
+
+config INFINIBAND_IPOIB_DEBUG_DATA
+	bool "IP-over-InfiniBand data path debugging"
+	depends on INFINIBAND_IPOIB_DEBUG
+	---help---
+	  This option compiles debugging code into the data path
+	  of the IPoIB driver.  The output can be turned on via the
+	  data_debug_level module parameter; however, even with output
+	  turned off, this debugging code will have some performance
+	  impact.

kernel/drivers/infiniband/ulp/ipoib/Makefile

@@ -0,0 +1,11 @@
+obj-$(CONFIG_INFINIBAND_IPOIB)			+= ib_ipoib.o
+
+ib_ipoib-y					:= ipoib_main.o \
+						   ipoib_ib.o \
+						   ipoib_multicast.o \
+						   ipoib_verbs.o \
+						   ipoib_vlan.o \
+						   ipoib_ethtool.o
+ib_ipoib-$(CONFIG_INFINIBAND_IPOIB_CM)		+= ipoib_cm.o
+ib_ipoib-$(CONFIG_INFINIBAND_IPOIB_DEBUG)	+= ipoib_fs.o
+

kernel/drivers/infiniband/ulp/ipoib/ipoib.h

@@ -0,0 +1,737 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2004 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.
+ */
+
+#ifndef _IPOIB_H
+#define _IPOIB_H
+
+#include <linux/list.h>
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include <linux/workqueue.h>
+#include <linux/kref.h>
+#include <linux/if_infiniband.h>
+#include <linux/mutex.h>
+
+#include <net/neighbour.h>
+
+#include <asm/atomic.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_pack.h>
+#include <rdma/ib_sa.h>
+#include <linux/inet_lro.h>
+
+/* constants */
+
+enum ipoib_flush_level {
+	IPOIB_FLUSH_LIGHT,
+	IPOIB_FLUSH_NORMAL,
+	IPOIB_FLUSH_HEAVY
+};
+
+enum {
+	IPOIB_ENCAP_LEN		  = 4,
+
+	IPOIB_UD_HEAD_SIZE	  = IB_GRH_BYTES + IPOIB_ENCAP_LEN,
+	IPOIB_UD_RX_SG		  = 2, /* max buffer needed for 4K mtu */
+
+	IPOIB_CM_MTU		  = 0x10000 - 0x10, /* padding to align header to 16 */
+	IPOIB_CM_BUF_SIZE	  = IPOIB_CM_MTU  + IPOIB_ENCAP_LEN,
+	IPOIB_CM_HEAD_SIZE	  = IPOIB_CM_BUF_SIZE % PAGE_SIZE,
+	IPOIB_CM_RX_SG		  = ALIGN(IPOIB_CM_BUF_SIZE, PAGE_SIZE) / PAGE_SIZE,
+	IPOIB_RX_RING_SIZE	  = 256,
+	IPOIB_TX_RING_SIZE	  = 128,
+	IPOIB_MAX_QUEUE_SIZE	  = 8192,
+	IPOIB_MIN_QUEUE_SIZE	  = 2,
+	IPOIB_CM_MAX_CONN_QP	  = 4096,
+
+	IPOIB_NUM_WC		  = 4,
+
+	IPOIB_MAX_PATH_REC_QUEUE  = 3,
+	IPOIB_MAX_MCAST_QUEUE	  = 3,
+
+	IPOIB_FLAG_OPER_UP	  = 0,
+	IPOIB_FLAG_INITIALIZED	  = 1,
+	IPOIB_FLAG_ADMIN_UP	  = 2,
+	IPOIB_PKEY_ASSIGNED	  = 3,
+	IPOIB_PKEY_STOP		  = 4,
+	IPOIB_FLAG_SUBINTERFACE	  = 5,
+	IPOIB_MCAST_RUN		  = 6,
+	IPOIB_STOP_REAPER	  = 7,
+	IPOIB_FLAG_ADMIN_CM	  = 9,
+	IPOIB_FLAG_UMCAST	  = 10,
+	IPOIB_FLAG_CSUM		  = 11,
+
+	IPOIB_MAX_BACKOFF_SECONDS = 16,
+
+	IPOIB_MCAST_FLAG_FOUND	  = 0,	/* used in set_multicast_list */
+	IPOIB_MCAST_FLAG_SENDONLY = 1,
+	IPOIB_MCAST_FLAG_BUSY	  = 2,	/* joining or already joined */
+	IPOIB_MCAST_FLAG_ATTACHED = 3,
+
+	IPOIB_MAX_LRO_DESCRIPTORS = 8,
+	IPOIB_LRO_MAX_AGGR 	  = 64,
+
+	MAX_SEND_CQE		  = 16,
+	IPOIB_CM_COPYBREAK	  = 256,
+};
+
+#define	IPOIB_OP_RECV   (1ul << 31)
+#ifdef CONFIG_INFINIBAND_IPOIB_CM
+#define	IPOIB_OP_CM     (1ul << 30)
+#else
+#define	IPOIB_OP_CM     (0)
+#endif
+
+/* structs */
+
+struct ipoib_header {
+	__be16	proto;
+	u16	reserved;
+};
+
+struct ipoib_pseudoheader {
+	u8  hwaddr[INFINIBAND_ALEN];
+};
+
+/* Used for all multicast joins (broadcast, IPv4 mcast and IPv6 mcast) */
+struct ipoib_mcast {
+	struct ib_sa_mcmember_rec mcmember;
+	struct ib_sa_multicast	 *mc;
+	struct ipoib_ah		 *ah;
+
+	struct rb_node    rb_node;
+	struct list_head  list;
+
+	unsigned long created;
+	unsigned long backoff;
+
+	unsigned long flags;
+	unsigned char logcount;
+
+	struct list_head  neigh_list;
+
+	struct sk_buff_head pkt_queue;
+
+	struct net_device *dev;
+};
+
+struct ipoib_rx_buf {
+	struct sk_buff *skb;
+	u64		mapping[IPOIB_UD_RX_SG];
+};
+
+struct ipoib_tx_buf {
+	struct sk_buff *skb;
+	u64		mapping[MAX_SKB_FRAGS + 1];
+};
+
+struct ipoib_cm_tx_buf {
+	struct sk_buff *skb;
+	u64		mapping;
+};
+
+struct ib_cm_id;
+
+struct ipoib_cm_data {
+	__be32 qpn; /* High byte MUST be ignored on receive */
+	__be32 mtu;
+};
+
+/*
+ * Quoting 10.3.1 Queue Pair and EE Context States:
+ *
+ * Note, for QPs that are associated with an SRQ, the Consumer should take the
+ * QP through the Error State before invoking a Destroy QP or a Modify QP to the
+ * Reset State.  The Consumer may invoke the Destroy QP without first performing
+ * a Modify QP to the Error State and waiting for the Affiliated Asynchronous
+ * Last WQE Reached Event. However, if the Consumer does not wait for the
+ * Affiliated Asynchronous Last WQE Reached Event, then WQE and Data Segment
+ * leakage may occur. Therefore, it is good programming practice to tear down a
+ * QP that is associated with an SRQ by using the following process:
+ *
+ * - Put the QP in the Error State
+ * - Wait for the Affiliated Asynchronous Last WQE Reached Event;
+ * - either:
+ *       drain the CQ by invoking the Poll CQ verb and either wait for CQ
+ *       to be empty or the number of Poll CQ operations has exceeded
+ *       CQ capacity size;
+ * - or
+ *       post another WR that completes on the same CQ and wait for this
+ *       WR to return as a WC;
+ * - and then invoke a Destroy QP or Reset QP.
+ *
+ * We use the second option and wait for a completion on the
+ * same CQ before destroying QPs attached to our SRQ.
+ */
+
+enum ipoib_cm_state {
+	IPOIB_CM_RX_LIVE,
+	IPOIB_CM_RX_ERROR, /* Ignored by stale task */
+	IPOIB_CM_RX_FLUSH  /* Last WQE Reached event observed */
+};
+
+struct ipoib_cm_rx {
+	struct ib_cm_id	       *id;
+	struct ib_qp	       *qp;
+	struct ipoib_cm_rx_buf *rx_ring;
+	struct list_head	list;
+	struct net_device      *dev;
+	unsigned long		jiffies;
+	enum ipoib_cm_state	state;
+	int			recv_count;
+};
+
+struct ipoib_cm_tx {
+	struct ib_cm_id	    *id;
+	struct ib_qp	    *qp;
+	struct list_head     list;
+	struct net_device   *dev;
+	struct ipoib_neigh  *neigh;
+	struct ipoib_path   *path;
+	struct ipoib_cm_tx_buf *tx_ring;
+	unsigned	     tx_head;
+	unsigned	     tx_tail;
+	unsigned long	     flags;
+	u32		     mtu;
+};
+
+struct ipoib_cm_rx_buf {
+	struct sk_buff *skb;
+	u64 mapping[IPOIB_CM_RX_SG];
+};
+
+struct ipoib_cm_dev_priv {
+	struct ib_srq	       *srq;
+	struct ipoib_cm_rx_buf *srq_ring;
+	struct ib_cm_id	       *id;
+	struct list_head	passive_ids;   /* state: LIVE */
+	struct list_head	rx_error_list; /* state: ERROR */
+	struct list_head	rx_flush_list; /* state: FLUSH, drain not started */
+	struct list_head	rx_drain_list; /* state: FLUSH, drain started */
+	struct list_head	rx_reap_list;  /* state: FLUSH, drain done */
+	struct work_struct      start_task;
+	struct work_struct      reap_task;
+	struct work_struct      skb_task;
+	struct work_struct      rx_reap_task;
+	struct delayed_work     stale_task;
+	struct sk_buff_head     skb_queue;
+	struct list_head	start_list;
+	struct list_head	reap_list;
+	struct ib_wc		ibwc[IPOIB_NUM_WC];
+	struct ib_sge		rx_sge[IPOIB_CM_RX_SG];
+	struct ib_recv_wr       rx_wr;
+	int			nonsrq_conn_qp;
+	int			max_cm_mtu;
+	int			num_frags;
+};
+
+struct ipoib_ethtool_st {
+	u16     coalesce_usecs;
+	u16     max_coalesced_frames;
+};
+
+struct ipoib_lro {
+	struct net_lro_mgr lro_mgr;
+	struct net_lro_desc lro_desc[IPOIB_MAX_LRO_DESCRIPTORS];
+};
+
+/*
+ * Device private locking: network stack tx_lock protects members used
+ * in TX fast path, lock protects everything else.  lock nests inside
+ * of tx_lock (ie tx_lock must be acquired first if needed).
+ */
+struct ipoib_dev_priv {
+	spinlock_t lock;
+
+	struct net_device *dev;
+
+	struct napi_struct napi;
+
+	unsigned long flags;
+
+	struct mutex vlan_mutex;
+
+	struct rb_root  path_tree;
+	struct list_head path_list;
+
+	struct ipoib_mcast *broadcast;
+	struct list_head multicast_list;
+	struct rb_root multicast_tree;
+
+	struct delayed_work pkey_poll_task;
+	struct delayed_work mcast_task;
+	struct work_struct carrier_on_task;
+	struct work_struct flush_light;
+	struct work_struct flush_normal;
+	struct work_struct flush_heavy;
+	struct work_struct restart_task;
+	struct delayed_work ah_reap_task;
+
+	struct ib_device *ca;
+	u8		  port;
+	u16		  pkey;
+	u16		  pkey_index;
+	struct ib_pd	 *pd;
+	struct ib_mr	 *mr;
+	struct ib_cq	 *recv_cq;
+	struct ib_cq	 *send_cq;
+	struct ib_qp	 *qp;
+	u32		  qkey;
+
+	union ib_gid local_gid;
+	u16	     local_lid;
+
+	unsigned int admin_mtu;
+	unsigned int mcast_mtu;
+	unsigned int max_ib_mtu;
+
+	struct ipoib_rx_buf *rx_ring;
+
+	struct ipoib_tx_buf *tx_ring;
+	unsigned	     tx_head;
+	unsigned	     tx_tail;
+	struct ib_sge	     tx_sge[MAX_SKB_FRAGS + 1];
+	struct ib_send_wr    tx_wr;
+	unsigned	     tx_outstanding;
+	struct ib_wc	     send_wc[MAX_SEND_CQE];
+
+	struct ib_recv_wr    rx_wr;
+	struct ib_sge	     rx_sge[IPOIB_UD_RX_SG];
+
+	struct ib_wc ibwc[IPOIB_NUM_WC];
+
+	struct list_head dead_ahs;
+
+	struct ib_event_handler event_handler;
+
+	struct net_device *parent;
+	struct list_head child_intfs;
+	struct list_head list;
+
+#ifdef CONFIG_INFINIBAND_IPOIB_CM
+	struct ipoib_cm_dev_priv cm;
+#endif
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+	struct list_head fs_list;
+	struct dentry *mcg_dentry;
+	struct dentry *path_dentry;
+#endif
+	int	hca_caps;
+	struct ipoib_ethtool_st ethtool;
+	struct timer_list poll_timer;
+
+	struct ipoib_lro lro;
+};
+
+struct ipoib_ah {
+	struct net_device *dev;
+	struct ib_ah	  *ah;
+	struct list_head   list;
+	struct kref	   ref;
+	unsigned	   last_send;
+};
+
+struct ipoib_path {
+	struct net_device    *dev;
+	struct ib_sa_path_rec pathrec;
+	struct ipoib_ah      *ah;
+	struct sk_buff_head   queue;
+
+	struct list_head      neigh_list;
+
+	int		      query_id;
+	struct ib_sa_query   *query;
+	struct completion     done;
+
+	struct rb_node	      rb_node;
+	struct list_head      list;
+	int  		      valid;
+};
+
+struct ipoib_neigh {
+	struct ipoib_ah    *ah;
+#ifdef CONFIG_INFINIBAND_IPOIB_CM
+	struct ipoib_cm_tx *cm;
+#endif
+	union ib_gid	    dgid;
+	struct sk_buff_head queue;
+
+	struct neighbour   *neighbour;
+	struct net_device *dev;
+
+	struct list_head    list;
+};
+
+#define IPOIB_UD_MTU(ib_mtu)		(ib_mtu - IPOIB_ENCAP_LEN)
+#define IPOIB_UD_BUF_SIZE(ib_mtu)	(ib_mtu + IB_GRH_BYTES)
+
+static inline int ipoib_ud_need_sg(unsigned int ib_mtu)
+{
+	return IPOIB_UD_BUF_SIZE(ib_mtu) > PAGE_SIZE;
+}
+
+/*
+ * We stash a pointer to our private neighbour information after our
+ * hardware address in neigh->ha.  The ALIGN() expression here makes
+ * sure that this pointer is stored aligned so that an unaligned
+ * load is not needed to dereference it.
+ */
+static inline struct ipoib_neigh **to_ipoib_neigh(struct neighbour *neigh)
+{
+	return (void*) neigh + ALIGN(offsetof(struct neighbour, ha) +
+				     INFINIBAND_ALEN, sizeof(void *));
+}
+
+struct ipoib_neigh *ipoib_neigh_alloc(struct neighbour *neigh,
+				      struct net_device *dev);
+void ipoib_neigh_free(struct net_device *dev, struct ipoib_neigh *neigh);
+
+extern struct workqueue_struct *ipoib_workqueue;
+
+/* functions */
+
+int ipoib_poll(struct napi_struct *napi, int budget);
+void ipoib_ib_completion(struct ib_cq *cq, void *dev_ptr);
+void ipoib_send_comp_handler(struct ib_cq *cq, void *dev_ptr);
+
+struct ipoib_ah *ipoib_create_ah(struct net_device *dev,
+				 struct ib_pd *pd, struct ib_ah_attr *attr);
+void ipoib_free_ah(struct kref *kref);
+static inline void ipoib_put_ah(struct ipoib_ah *ah)
+{
+	kref_put(&ah->ref, ipoib_free_ah);
+}
+
+int ipoib_open(struct net_device *dev);
+int ipoib_add_pkey_attr(struct net_device *dev);
+int ipoib_add_umcast_attr(struct net_device *dev);
+
+void ipoib_send(struct net_device *dev, struct sk_buff *skb,
+		struct ipoib_ah *address, u32 qpn);
+void ipoib_reap_ah(struct work_struct *work);
+
+void ipoib_mark_paths_invalid(struct net_device *dev);
+void ipoib_flush_paths(struct net_device *dev);
+struct ipoib_dev_priv *ipoib_intf_alloc(const char *format);
+
+int ipoib_ib_dev_init(struct net_device *dev, struct ib_device *ca, int port);
+void ipoib_ib_dev_flush_light(struct work_struct *work);
+void ipoib_ib_dev_flush_normal(struct work_struct *work);
+void ipoib_ib_dev_flush_heavy(struct work_struct *work);
+void ipoib_pkey_event(struct work_struct *work);
+void ipoib_ib_dev_cleanup(struct net_device *dev);
+
+int ipoib_ib_dev_open(struct net_device *dev);
+int ipoib_ib_dev_up(struct net_device *dev);
+int ipoib_ib_dev_down(struct net_device *dev, int flush);
+int ipoib_ib_dev_stop(struct net_device *dev, int flush);
+
+int ipoib_dev_init(struct net_device *dev, struct ib_device *ca, int port);
+void ipoib_dev_cleanup(struct net_device *dev);
+
+void ipoib_mcast_join_task(struct work_struct *work);
+void ipoib_mcast_carrier_on_task(struct work_struct *work);
+void ipoib_mcast_send(struct net_device *dev, void *mgid, struct sk_buff *skb);
+
+void ipoib_mcast_restart_task(struct work_struct *work);
+int ipoib_mcast_start_thread(struct net_device *dev);
+int ipoib_mcast_stop_thread(struct net_device *dev, int flush);
+
+void ipoib_mcast_dev_down(struct net_device *dev);
+void ipoib_mcast_dev_flush(struct net_device *dev);
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+struct ipoib_mcast_iter *ipoib_mcast_iter_init(struct net_device *dev);
+int ipoib_mcast_iter_next(struct ipoib_mcast_iter *iter);
+void ipoib_mcast_iter_read(struct ipoib_mcast_iter *iter,
+				  union ib_gid *gid,
+				  unsigned long *created,
+				  unsigned int *queuelen,
+				  unsigned int *complete,
+				  unsigned int *send_only);
+
+struct ipoib_path_iter *ipoib_path_iter_init(struct net_device *dev);
+int ipoib_path_iter_next(struct ipoib_path_iter *iter);
+void ipoib_path_iter_read(struct ipoib_path_iter *iter,
+			  struct ipoib_path *path);
+#endif
+
+int ipoib_mcast_attach(struct net_device *dev, u16 mlid,
+		       union ib_gid *mgid, int set_qkey);
+
+int ipoib_init_qp(struct net_device *dev);
+int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca);
+void ipoib_transport_dev_cleanup(struct net_device *dev);
+
+void ipoib_event(struct ib_event_handler *handler,
+		 struct ib_event *record);
+
+int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey);
+int ipoib_vlan_delete(struct net_device *pdev, unsigned short pkey);
+
+void ipoib_pkey_poll(struct work_struct *work);
+int ipoib_pkey_dev_delay_open(struct net_device *dev);
+void ipoib_drain_cq(struct net_device *dev);
+
+void ipoib_set_ethtool_ops(struct net_device *dev);
+int ipoib_set_dev_features(struct ipoib_dev_priv *priv, struct ib_device *hca);
+
+#ifdef CONFIG_INFINIBAND_IPOIB_CM
+
+#define IPOIB_FLAGS_RC		0x80
+#define IPOIB_FLAGS_UC		0x40
+
+/* We don't support UC connections at the moment */
+#define IPOIB_CM_SUPPORTED(ha)   (ha[0] & (IPOIB_FLAGS_RC))
+
+extern int ipoib_max_conn_qp;
+
+static inline int ipoib_cm_admin_enabled(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	return IPOIB_CM_SUPPORTED(dev->dev_addr) &&
+		test_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags);
+}
+
+static inline int ipoib_cm_enabled(struct net_device *dev, struct neighbour *n)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	return IPOIB_CM_SUPPORTED(n->ha) &&
+		test_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags);
+}
+
+static inline int ipoib_cm_up(struct ipoib_neigh *neigh)
+
+{
+	return test_bit(IPOIB_FLAG_OPER_UP, &neigh->cm->flags);
+}
+
+static inline struct ipoib_cm_tx *ipoib_cm_get(struct ipoib_neigh *neigh)
+{
+	return neigh->cm;
+}
+
+static inline void ipoib_cm_set(struct ipoib_neigh *neigh, struct ipoib_cm_tx *tx)
+{
+	neigh->cm = tx;
+}
+
+static inline int ipoib_cm_has_srq(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	return !!priv->cm.srq;
+}
+
+static inline unsigned int ipoib_cm_max_mtu(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	return priv->cm.max_cm_mtu;
+}
+
+void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_tx *tx);
+int ipoib_cm_dev_open(struct net_device *dev);
+void ipoib_cm_dev_stop(struct net_device *dev);
+int ipoib_cm_dev_init(struct net_device *dev);
+int ipoib_cm_add_mode_attr(struct net_device *dev);
+void ipoib_cm_dev_cleanup(struct net_device *dev);
+struct ipoib_cm_tx *ipoib_cm_create_tx(struct net_device *dev, struct ipoib_path *path,
+				    struct ipoib_neigh *neigh);
+void ipoib_cm_destroy_tx(struct ipoib_cm_tx *tx);
+void ipoib_cm_skb_too_long(struct net_device *dev, struct sk_buff *skb,
+			   unsigned int mtu);
+void ipoib_cm_handle_rx_wc(struct net_device *dev, struct ib_wc *wc);
+void ipoib_cm_handle_tx_wc(struct net_device *dev, struct ib_wc *wc);
+#else
+
+struct ipoib_cm_tx;
+
+#define ipoib_max_conn_qp 0
+
+static inline int ipoib_cm_admin_enabled(struct net_device *dev)
+{
+	return 0;
+}
+static inline int ipoib_cm_enabled(struct net_device *dev, struct neighbour *n)
+
+{
+	return 0;
+}
+
+static inline int ipoib_cm_up(struct ipoib_neigh *neigh)
+
+{
+	return 0;
+}
+
+static inline struct ipoib_cm_tx *ipoib_cm_get(struct ipoib_neigh *neigh)
+{
+	return NULL;
+}
+
+static inline void ipoib_cm_set(struct ipoib_neigh *neigh, struct ipoib_cm_tx *tx)
+{
+}
+
+static inline int ipoib_cm_has_srq(struct net_device *dev)
+{
+	return 0;
+}
+
+static inline unsigned int ipoib_cm_max_mtu(struct net_device *dev)
+{
+	return 0;
+}
+
+static inline
+void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_tx *tx)
+{
+	return;
+}
+
+static inline
+int ipoib_cm_dev_open(struct net_device *dev)
+{
+	return 0;
+}
+
+static inline
+void ipoib_cm_dev_stop(struct net_device *dev)
+{
+	return;
+}
+
+static inline
+int ipoib_cm_dev_init(struct net_device *dev)
+{
+	return -ENOSYS;
+}
+
+static inline
+void ipoib_cm_dev_cleanup(struct net_device *dev)
+{
+	return;
+}
+
+static inline
+struct ipoib_cm_tx *ipoib_cm_create_tx(struct net_device *dev, struct ipoib_path *path,
+				    struct ipoib_neigh *neigh)
+{
+	return NULL;
+}
+
+static inline
+void ipoib_cm_destroy_tx(struct ipoib_cm_tx *tx)
+{
+	return;
+}
+
+static inline
+int ipoib_cm_add_mode_attr(struct net_device *dev)
+{
+	return 0;
+}
+
+static inline void ipoib_cm_skb_too_long(struct net_device *dev, struct sk_buff *skb,
+					 unsigned int mtu)
+{
+	dev_kfree_skb_any(skb);
+}
+
+static inline void ipoib_cm_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
+{
+}
+
+static inline void ipoib_cm_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
+{
+}
+#endif
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+void ipoib_create_debug_files(struct net_device *dev);
+void ipoib_delete_debug_files(struct net_device *dev);
+int ipoib_register_debugfs(void);
+void ipoib_unregister_debugfs(void);
+#else
+static inline void ipoib_create_debug_files(struct net_device *dev) { }
+static inline void ipoib_delete_debug_files(struct net_device *dev) { }
+static inline int ipoib_register_debugfs(void) { return 0; }
+static inline void ipoib_unregister_debugfs(void) { }
+#endif
+
+#define ipoib_printk(level, priv, format, arg...)	\
+	printk(level "%s: " format, ((struct ipoib_dev_priv *) priv)->dev->name , ## arg)
+#define ipoib_warn(priv, format, arg...)		\
+	ipoib_printk(KERN_WARNING, priv, format , ## arg)
+
+extern int ipoib_sendq_size;
+extern int ipoib_recvq_size;
+
+extern struct ib_sa_client ipoib_sa_client;
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+extern int ipoib_debug_level;
+
+#define ipoib_dbg(priv, format, arg...)			\
+	do {						\
+		if (ipoib_debug_level > 0)			\
+			ipoib_printk(KERN_DEBUG, priv, format , ## arg); \
+	} while (0)
+#define ipoib_dbg_mcast(priv, format, arg...)		\
+	do {						\
+		if (mcast_debug_level > 0)		\
+			ipoib_printk(KERN_DEBUG, priv, format , ## arg); \
+	} while (0)
+#else /* CONFIG_INFINIBAND_IPOIB_DEBUG */
+#define ipoib_dbg(priv, format, arg...)			\
+	do { (void) (priv); } while (0)
+#define ipoib_dbg_mcast(priv, format, arg...)		\
+	do { (void) (priv); } while (0)
+#endif /* CONFIG_INFINIBAND_IPOIB_DEBUG */
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG_DATA
+#define ipoib_dbg_data(priv, format, arg...)		\
+	do {						\
+		if (data_debug_level > 0)		\
+			ipoib_printk(KERN_DEBUG, priv, format , ## arg); \
+	} while (0)
+#else /* CONFIG_INFINIBAND_IPOIB_DEBUG_DATA */
+#define ipoib_dbg_data(priv, format, arg...)		\
+	do { (void) (priv); } while (0)
+#endif /* CONFIG_INFINIBAND_IPOIB_DEBUG_DATA */
+
+#define IPOIB_QPN(ha) (be32_to_cpup((__be32 *) ha) & 0xffffff)
+
+#endif /* _IPOIB_H */

kernel/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c

@@ -0,0 +1,152 @@
+/*
+ * Copyright (c) 2007 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.
+ */
+
+#include <linux/kernel.h>
+#include <linux/ethtool.h>
+#include <linux/netdevice.h>
+
+#include "ipoib.h"
+
+static void ipoib_get_drvinfo(struct net_device *netdev,
+			      struct ethtool_drvinfo *drvinfo)
+{
+	strncpy(drvinfo->driver, "ipoib", sizeof(drvinfo->driver) - 1);
+}
+
+static u32 ipoib_get_rx_csum(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	return test_bit(IPOIB_FLAG_CSUM, &priv->flags) &&
+		!test_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags);
+}
+
+static int ipoib_get_coalesce(struct net_device *dev,
+			      struct ethtool_coalesce *coal)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	coal->rx_coalesce_usecs = priv->ethtool.coalesce_usecs;
+	coal->tx_coalesce_usecs = priv->ethtool.coalesce_usecs;
+	coal->rx_max_coalesced_frames = priv->ethtool.max_coalesced_frames;
+	coal->tx_max_coalesced_frames = priv->ethtool.max_coalesced_frames;
+
+	return 0;
+}
+
+static int ipoib_set_coalesce(struct net_device *dev,
+			      struct ethtool_coalesce *coal)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	int ret;
+
+	/*
+	 * Since IPoIB uses a single CQ for both rx and tx, we assume
+	 * that rx params dictate the configuration.  These values are
+	 * saved in the private data and returned when ipoib_get_coalesce()
+	 * is called.
+	 */
+	if (coal->rx_coalesce_usecs       > 0xffff ||
+	    coal->rx_max_coalesced_frames > 0xffff)
+		return -EINVAL;
+
+	ret = ib_modify_cq(priv->recv_cq, coal->rx_max_coalesced_frames,
+			   coal->rx_coalesce_usecs);
+	if (ret && ret != -ENOSYS) {
+		ipoib_warn(priv, "failed modifying CQ (%d)\n", ret);
+		return ret;
+	}
+
+	coal->tx_coalesce_usecs       = coal->rx_coalesce_usecs;
+	coal->tx_max_coalesced_frames = coal->rx_max_coalesced_frames;
+	priv->ethtool.coalesce_usecs       = coal->rx_coalesce_usecs;
+	priv->ethtool.max_coalesced_frames = coal->rx_max_coalesced_frames;
+
+	return 0;
+}
+
+static const char ipoib_stats_keys[][ETH_GSTRING_LEN] = {
+	"LRO aggregated", "LRO flushed",
+	"LRO avg aggr", "LRO no desc"
+};
+
+static void ipoib_get_strings(struct net_device *netdev, u32 stringset, u8 *data)
+{
+	switch (stringset) {
+	case ETH_SS_STATS:
+		memcpy(data, *ipoib_stats_keys,	sizeof(ipoib_stats_keys));
+		break;
+	}
+}
+
+static int ipoib_get_sset_count(struct net_device *dev, int sset)
+{
+	switch (sset) {
+	case ETH_SS_STATS:
+		return ARRAY_SIZE(ipoib_stats_keys);
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
+static void ipoib_get_ethtool_stats(struct net_device *dev,
+				struct ethtool_stats *stats, uint64_t *data)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	int index = 0;
+
+	/* Get LRO statistics */
+	data[index++] = priv->lro.lro_mgr.stats.aggregated;
+	data[index++] = priv->lro.lro_mgr.stats.flushed;
+	if (priv->lro.lro_mgr.stats.flushed)
+		data[index++] = priv->lro.lro_mgr.stats.aggregated /
+				priv->lro.lro_mgr.stats.flushed;
+	else
+		data[index++] = 0;
+	data[index++] = priv->lro.lro_mgr.stats.no_desc;
+}
+
+static const struct ethtool_ops ipoib_ethtool_ops = {
+	.get_drvinfo		= ipoib_get_drvinfo,
+	.get_rx_csum		= ipoib_get_rx_csum,
+	.get_coalesce		= ipoib_get_coalesce,
+	.set_coalesce		= ipoib_set_coalesce,
+	.get_flags		= ethtool_op_get_flags,
+	.set_flags		= ethtool_op_set_flags,
+	.get_strings		= ipoib_get_strings,
+	.get_sset_count		= ipoib_get_sset_count,
+	.get_ethtool_stats	= ipoib_get_ethtool_stats,
+};
+
+void ipoib_set_ethtool_ops(struct net_device *dev)
+{
+	SET_ETHTOOL_OPS(dev, &ipoib_ethtool_ops);
+}

kernel/drivers/infiniband/ulp/ipoib/ipoib_fs.c

@@ -0,0 +1,298 @@
+/*
+ * 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.
+ */
+
+#include <linux/err.h>
+#include <linux/seq_file.h>
+
+struct file_operations;
+
+#include <linux/debugfs.h>
+
+#include "ipoib.h"
+
+static struct dentry *ipoib_root;
+
+static void format_gid(union ib_gid *gid, char *buf)
+{
+	int i, n;
+
+	for (n = 0, i = 0; i < 8; ++i) {
+		n += sprintf(buf + n, "%x",
+			     be16_to_cpu(((__be16 *) gid->raw)[i]));
+		if (i < 7)
+			buf[n++] = ':';
+	}
+}
+
+static void *ipoib_mcg_seq_start(struct seq_file *file, loff_t *pos)
+{
+	struct ipoib_mcast_iter *iter;
+	loff_t n = *pos;
+
+	iter = ipoib_mcast_iter_init(file->private);
+	if (!iter)
+		return NULL;
+
+	while (n--) {
+		if (ipoib_mcast_iter_next(iter)) {
+			kfree(iter);
+			return NULL;
+		}
+	}
+
+	return iter;
+}
+
+static void *ipoib_mcg_seq_next(struct seq_file *file, void *iter_ptr,
+				   loff_t *pos)
+{
+	struct ipoib_mcast_iter *iter = iter_ptr;
+
+	(*pos)++;
+
+	if (ipoib_mcast_iter_next(iter)) {
+		kfree(iter);
+		return NULL;
+	}
+
+	return iter;
+}
+
+static void ipoib_mcg_seq_stop(struct seq_file *file, void *iter_ptr)
+{
+	/* nothing for now */
+}
+
+static int ipoib_mcg_seq_show(struct seq_file *file, void *iter_ptr)
+{
+	struct ipoib_mcast_iter *iter = iter_ptr;
+	char gid_buf[sizeof "ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"];
+	union ib_gid mgid;
+	unsigned long created;
+	unsigned int queuelen, complete, send_only;
+
+	if (!iter)
+		return 0;
+
+	ipoib_mcast_iter_read(iter, &mgid, &created, &queuelen,
+			      &complete, &send_only);
+
+	format_gid(&mgid, gid_buf);
+
+	seq_printf(file,
+		   "GID: %s\n"
+		   "  created: %10ld\n"
+		   "  queuelen: %9d\n"
+		   "  complete: %9s\n"
+		   "  send_only: %8s\n"
+		   "\n",
+		   gid_buf, created, queuelen,
+		   complete ? "yes" : "no",
+		   send_only ? "yes" : "no");
+
+	return 0;
+}
+
+static const struct seq_operations ipoib_mcg_seq_ops = {
+	.start = ipoib_mcg_seq_start,
+	.next  = ipoib_mcg_seq_next,
+	.stop  = ipoib_mcg_seq_stop,
+	.show  = ipoib_mcg_seq_show,
+};
+
+static int ipoib_mcg_open(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq;
+	int ret;
+
+	ret = seq_open(file, &ipoib_mcg_seq_ops);
+	if (ret)
+		return ret;
+
+	seq = file->private_data;
+	seq->private = inode->i_private;
+
+	return 0;
+}
+
+static const struct file_operations ipoib_mcg_fops = {
+	.owner   = THIS_MODULE,
+	.open    = ipoib_mcg_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release
+};
+
+static void *ipoib_path_seq_start(struct seq_file *file, loff_t *pos)
+{
+	struct ipoib_path_iter *iter;
+	loff_t n = *pos;
+
+	iter = ipoib_path_iter_init(file->private);
+	if (!iter)
+		return NULL;
+
+	while (n--) {
+		if (ipoib_path_iter_next(iter)) {
+			kfree(iter);
+			return NULL;
+		}
+	}
+
+	return iter;
+}
+
+static void *ipoib_path_seq_next(struct seq_file *file, void *iter_ptr,
+				   loff_t *pos)
+{
+	struct ipoib_path_iter *iter = iter_ptr;
+
+	(*pos)++;
+
+	if (ipoib_path_iter_next(iter)) {
+		kfree(iter);
+		return NULL;
+	}
+
+	return iter;
+}
+
+static void ipoib_path_seq_stop(struct seq_file *file, void *iter_ptr)
+{
+	/* nothing for now */
+}
+
+static int ipoib_path_seq_show(struct seq_file *file, void *iter_ptr)
+{
+	struct ipoib_path_iter *iter = iter_ptr;
+	char gid_buf[sizeof "ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"];
+	struct ipoib_path path;
+	int rate;
+
+	if (!iter)
+		return 0;
+
+	ipoib_path_iter_read(iter, &path);
+
+	format_gid(&path.pathrec.dgid, gid_buf);
+
+	seq_printf(file,
+		   "GID: %s\n"
+		   "  complete: %6s\n",
+		   gid_buf, path.pathrec.dlid ? "yes" : "no");
+
+	if (path.pathrec.dlid) {
+		rate = ib_rate_to_mult(path.pathrec.rate) * 25;
+
+		seq_printf(file,
+			   "  DLID:     0x%04x\n"
+			   "  SL: %12d\n"
+			   "  rate: %*d%s Gb/sec\n",
+			   be16_to_cpu(path.pathrec.dlid),
+			   path.pathrec.sl,
+			   10 - ((rate % 10) ? 2 : 0),
+			   rate / 10, rate % 10 ? ".5" : "");
+	}
+
+	seq_putc(file, '\n');
+
+	return 0;
+}
+
+static const struct seq_operations ipoib_path_seq_ops = {
+	.start = ipoib_path_seq_start,
+	.next  = ipoib_path_seq_next,
+	.stop  = ipoib_path_seq_stop,
+	.show  = ipoib_path_seq_show,
+};
+
+static int ipoib_path_open(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq;
+	int ret;
+
+	ret = seq_open(file, &ipoib_path_seq_ops);
+	if (ret)
+		return ret;
+
+	seq = file->private_data;
+	seq->private = inode->i_private;
+
+	return 0;
+}
+
+static const struct file_operations ipoib_path_fops = {
+	.owner   = THIS_MODULE,
+	.open    = ipoib_path_open,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.release = seq_release
+};
+
+void ipoib_create_debug_files(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	char name[IFNAMSIZ + sizeof "_path"];
+
+	snprintf(name, sizeof name, "%s_mcg", dev->name);
+	priv->mcg_dentry = debugfs_create_file(name, S_IFREG | S_IRUGO,
+					       ipoib_root, dev, &ipoib_mcg_fops);
+	if (!priv->mcg_dentry)
+		ipoib_warn(priv, "failed to create mcg debug file\n");
+
+	snprintf(name, sizeof name, "%s_path", dev->name);
+	priv->path_dentry = debugfs_create_file(name, S_IFREG | S_IRUGO,
+						ipoib_root, dev, &ipoib_path_fops);
+	if (!priv->path_dentry)
+		ipoib_warn(priv, "failed to create path debug file\n");
+}
+
+void ipoib_delete_debug_files(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	if (priv->mcg_dentry)
+		debugfs_remove(priv->mcg_dentry);
+	if (priv->path_dentry)
+		debugfs_remove(priv->path_dentry);
+}
+
+int ipoib_register_debugfs(void)
+{
+	ipoib_root = debugfs_create_dir("ipoib", NULL);
+	return ipoib_root ? 0 : -ENOMEM;
+}
+
+void ipoib_unregister_debugfs(void)
+{
+	debugfs_remove(ipoib_root);
+}

kernel/drivers/infiniband/ulp/ipoib/ipoib_multicast.c

@@ -0,0 +1,965 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2004 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.
+ */
+
+#include <linux/skbuff.h>
+#include <linux/rtnetlink.h>
+#include <linux/ip.h>
+#include <linux/in.h>
+#include <linux/igmp.h>
+#include <linux/inetdevice.h>
+#include <linux/delay.h>
+#include <linux/completion.h>
+
+#include <net/dst.h>
+
+#include "ipoib.h"
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+static int mcast_debug_level;
+
+module_param(mcast_debug_level, int, 0644);
+MODULE_PARM_DESC(mcast_debug_level,
+		 "Enable multicast debug tracing if > 0");
+#endif
+
+static DEFINE_MUTEX(mcast_mutex);
+
+struct ipoib_mcast_iter {
+	struct net_device *dev;
+	union ib_gid       mgid;
+	unsigned long      created;
+	unsigned int       queuelen;
+	unsigned int       complete;
+	unsigned int       send_only;
+};
+
+static void ipoib_mcast_free(struct ipoib_mcast *mcast)
+{
+	struct net_device *dev = mcast->dev;
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ipoib_neigh *neigh, *tmp;
+	int tx_dropped = 0;
+
+	ipoib_dbg_mcast(netdev_priv(dev), "deleting multicast group %pI6\n",
+			mcast->mcmember.mgid.raw);
+
+	spin_lock_irq(&priv->lock);
+
+	list_for_each_entry_safe(neigh, tmp, &mcast->neigh_list, list) {
+		/*
+		 * It's safe to call ipoib_put_ah() inside priv->lock
+		 * here, because we know that mcast->ah will always
+		 * hold one more reference, so ipoib_put_ah() will
+		 * never do more than decrement the ref count.
+		 */
+		if (neigh->ah)
+			ipoib_put_ah(neigh->ah);
+		ipoib_neigh_free(dev, neigh);
+	}
+
+	spin_unlock_irq(&priv->lock);
+
+	if (mcast->ah)
+		ipoib_put_ah(mcast->ah);
+
+	while (!skb_queue_empty(&mcast->pkt_queue)) {
+		++tx_dropped;
+		dev_kfree_skb_any(skb_dequeue(&mcast->pkt_queue));
+	}
+
+	netif_tx_lock_bh(dev);
+	dev->stats.tx_dropped += tx_dropped;
+	netif_tx_unlock_bh(dev);
+
+	kfree(mcast);
+}
+
+static struct ipoib_mcast *ipoib_mcast_alloc(struct net_device *dev,
+					     int can_sleep)
+{
+	struct ipoib_mcast *mcast;
+
+	mcast = kzalloc(sizeof *mcast, can_sleep ? GFP_KERNEL : GFP_ATOMIC);
+	if (!mcast)
+		return NULL;
+
+	mcast->dev = dev;
+	mcast->created = jiffies;
+	mcast->backoff = 1;
+
+	INIT_LIST_HEAD(&mcast->list);
+	INIT_LIST_HEAD(&mcast->neigh_list);
+	skb_queue_head_init(&mcast->pkt_queue);
+
+	return mcast;
+}
+
+static struct ipoib_mcast *__ipoib_mcast_find(struct net_device *dev, void *mgid)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct rb_node *n = priv->multicast_tree.rb_node;
+
+	while (n) {
+		struct ipoib_mcast *mcast;
+		int ret;
+
+		mcast = rb_entry(n, struct ipoib_mcast, rb_node);
+
+		ret = memcmp(mgid, mcast->mcmember.mgid.raw,
+			     sizeof (union ib_gid));
+		if (ret < 0)
+			n = n->rb_left;
+		else if (ret > 0)
+			n = n->rb_right;
+		else
+			return mcast;
+	}
+
+	return NULL;
+}
+
+static int __ipoib_mcast_add(struct net_device *dev, struct ipoib_mcast *mcast)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct rb_node **n = &priv->multicast_tree.rb_node, *pn = NULL;
+
+	while (*n) {
+		struct ipoib_mcast *tmcast;
+		int ret;
+
+		pn = *n;
+		tmcast = rb_entry(pn, struct ipoib_mcast, rb_node);
+
+		ret = memcmp(mcast->mcmember.mgid.raw, tmcast->mcmember.mgid.raw,
+			     sizeof (union ib_gid));
+		if (ret < 0)
+			n = &pn->rb_left;
+		else if (ret > 0)
+			n = &pn->rb_right;
+		else
+			return -EEXIST;
+	}
+
+	rb_link_node(&mcast->rb_node, pn, n);
+	rb_insert_color(&mcast->rb_node, &priv->multicast_tree);
+
+	return 0;
+}
+
+static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast,
+				   struct ib_sa_mcmember_rec *mcmember)
+{
+	struct net_device *dev = mcast->dev;
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ipoib_ah *ah;
+	int ret;
+	int set_qkey = 0;
+
+	mcast->mcmember = *mcmember;
+
+	/* Set the cached Q_Key before we attach if it's the broadcast group */
+	if (!memcmp(mcast->mcmember.mgid.raw, priv->dev->broadcast + 4,
+		    sizeof (union ib_gid))) {
+		spin_lock_irq(&priv->lock);
+		if (!priv->broadcast) {
+			spin_unlock_irq(&priv->lock);
+			return -EAGAIN;
+		}
+		priv->qkey = be32_to_cpu(priv->broadcast->mcmember.qkey);
+		spin_unlock_irq(&priv->lock);
+		priv->tx_wr.wr.ud.remote_qkey = priv->qkey;
+		set_qkey = 1;
+	}
+
+	if (!test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags)) {
+		if (test_and_set_bit(IPOIB_MCAST_FLAG_ATTACHED, &mcast->flags)) {
+			ipoib_warn(priv, "multicast group %pI6 already attached\n",
+				   mcast->mcmember.mgid.raw);
+
+			return 0;
+		}
+
+		ret = ipoib_mcast_attach(dev, be16_to_cpu(mcast->mcmember.mlid),
+					 &mcast->mcmember.mgid, set_qkey);
+		if (ret < 0) {
+			ipoib_warn(priv, "couldn't attach QP to multicast group %pI6\n",
+				   mcast->mcmember.mgid.raw);
+
+			clear_bit(IPOIB_MCAST_FLAG_ATTACHED, &mcast->flags);
+			return ret;
+		}
+	}
+
+	{
+		struct ib_ah_attr av = {
+			.dlid	       = be16_to_cpu(mcast->mcmember.mlid),
+			.port_num      = priv->port,
+			.sl	       = mcast->mcmember.sl,
+			.ah_flags      = IB_AH_GRH,
+			.static_rate   = mcast->mcmember.rate,
+			.grh	       = {
+				.flow_label    = be32_to_cpu(mcast->mcmember.flow_label),
+				.hop_limit     = mcast->mcmember.hop_limit,
+				.sgid_index    = 0,
+				.traffic_class = mcast->mcmember.traffic_class
+			}
+		};
+		av.grh.dgid = mcast->mcmember.mgid;
+
+		ah = ipoib_create_ah(dev, priv->pd, &av);
+		if (!ah) {
+			ipoib_warn(priv, "ib_address_create failed\n");
+		} else {
+			spin_lock_irq(&priv->lock);
+			mcast->ah = ah;
+			spin_unlock_irq(&priv->lock);
+
+			ipoib_dbg_mcast(priv, "MGID %pI6 AV %p, LID 0x%04x, SL %d\n",
+					mcast->mcmember.mgid.raw,
+					mcast->ah->ah,
+					be16_to_cpu(mcast->mcmember.mlid),
+					mcast->mcmember.sl);
+		}
+	}
+
+	/* actually send any queued packets */
+	netif_tx_lock_bh(dev);
+	while (!skb_queue_empty(&mcast->pkt_queue)) {
+		struct sk_buff *skb = skb_dequeue(&mcast->pkt_queue);
+		netif_tx_unlock_bh(dev);
+
+		skb->dev = dev;
+
+		if (!skb_dst(skb) || !skb_dst(skb)->neighbour) {
+			/* put pseudoheader back on for next time */
+			skb_push(skb, sizeof (struct ipoib_pseudoheader));
+		}
+
+		if (dev_queue_xmit(skb))
+			ipoib_warn(priv, "dev_queue_xmit failed to requeue packet\n");
+		netif_tx_lock_bh(dev);
+	}
+	netif_tx_unlock_bh(dev);
+
+	return 0;
+}
+
+static int
+ipoib_mcast_sendonly_join_complete(int status,
+				   struct ib_sa_multicast *multicast)
+{
+	struct ipoib_mcast *mcast = multicast->context;
+	struct net_device *dev = mcast->dev;
+
+	/* We trap for port events ourselves. */
+	if (status == -ENETRESET)
+		return 0;
+
+	if (!status)
+		status = ipoib_mcast_join_finish(mcast, &multicast->rec);
+
+	if (status) {
+		if (mcast->logcount++ < 20)
+			ipoib_dbg_mcast(netdev_priv(dev), "multicast join failed for %pI6, status %d\n",
+					mcast->mcmember.mgid.raw, status);
+
+		/* Flush out any queued packets */
+		netif_tx_lock_bh(dev);
+		while (!skb_queue_empty(&mcast->pkt_queue)) {
+			++dev->stats.tx_dropped;
+			dev_kfree_skb_any(skb_dequeue(&mcast->pkt_queue));
+		}
+		netif_tx_unlock_bh(dev);
+
+		/* Clear the busy flag so we try again */
+		status = test_and_clear_bit(IPOIB_MCAST_FLAG_BUSY,
+					    &mcast->flags);
+	}
+	return status;
+}
+
+static int ipoib_mcast_sendonly_join(struct ipoib_mcast *mcast)
+{
+	struct net_device *dev = mcast->dev;
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ib_sa_mcmember_rec rec = {
+#if 0				/* Some SMs don't support send-only yet */
+		.join_state = 4
+#else
+		.join_state = 1
+#endif
+	};
+	int ret = 0;
+
+	if (!test_bit(IPOIB_FLAG_OPER_UP, &priv->flags)) {
+		ipoib_dbg_mcast(priv, "device shutting down, no multicast joins\n");
+		return -ENODEV;
+	}
+
+	if (test_and_set_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags)) {
+		ipoib_dbg_mcast(priv, "multicast entry busy, skipping\n");
+		return -EBUSY;
+	}
+
+	rec.mgid     = mcast->mcmember.mgid;
+	rec.port_gid = priv->local_gid;
+	rec.pkey     = cpu_to_be16(priv->pkey);
+
+	mcast->mc = ib_sa_join_multicast(&ipoib_sa_client, priv->ca,
+					 priv->port, &rec,
+					 IB_SA_MCMEMBER_REC_MGID	|
+					 IB_SA_MCMEMBER_REC_PORT_GID	|
+					 IB_SA_MCMEMBER_REC_PKEY	|
+					 IB_SA_MCMEMBER_REC_JOIN_STATE,
+					 GFP_ATOMIC,
+					 ipoib_mcast_sendonly_join_complete,
+					 mcast);
+	if (IS_ERR(mcast->mc)) {
+		ret = PTR_ERR(mcast->mc);
+		clear_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags);
+		ipoib_warn(priv, "ib_sa_join_multicast failed (ret = %d)\n",
+			   ret);
+	} else {
+		ipoib_dbg_mcast(priv, "no multicast record for %pI6, starting join\n",
+				mcast->mcmember.mgid.raw);
+	}
+
+	return ret;
+}
+
+void ipoib_mcast_carrier_on_task(struct work_struct *work)
+{
+	struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv,
+						   carrier_on_task);
+	struct ib_port_attr attr;
+
+	/*
+	 * Take rtnl_lock to avoid racing with ipoib_stop() and
+	 * turning the carrier back on while a device is being
+	 * removed.
+	 */
+	if (ib_query_port(priv->ca, priv->port, &attr) ||
+	    attr.state != IB_PORT_ACTIVE) {
+		ipoib_dbg(priv, "Keeping carrier off until IB port is active\n");
+		return;
+	}
+
+	rtnl_lock();
+	netif_carrier_on(priv->dev);
+	rtnl_unlock();
+}
+
+static int ipoib_mcast_join_complete(int status,
+				     struct ib_sa_multicast *multicast)
+{
+	struct ipoib_mcast *mcast = multicast->context;
+	struct net_device *dev = mcast->dev;
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	ipoib_dbg_mcast(priv, "join completion for %pI6 (status %d)\n",
+			mcast->mcmember.mgid.raw, status);
+
+	/* We trap for port events ourselves. */
+	if (status == -ENETRESET)
+		return 0;
+
+	if (!status)
+		status = ipoib_mcast_join_finish(mcast, &multicast->rec);
+
+	if (!status) {
+		mcast->backoff = 1;
+		mutex_lock(&mcast_mutex);
+		if (test_bit(IPOIB_MCAST_RUN, &priv->flags))
+			queue_delayed_work(ipoib_workqueue,
+					   &priv->mcast_task, 0);
+		mutex_unlock(&mcast_mutex);
+
+		/*
+		 * Defer carrier on work to ipoib_workqueue to avoid a
+		 * deadlock on rtnl_lock here.
+		 */
+		if (mcast == priv->broadcast)
+			queue_work(ipoib_workqueue, &priv->carrier_on_task);
+
+		return 0;
+	}
+
+	if (mcast->logcount++ < 20) {
+		if (status == -ETIMEDOUT || status == -EAGAIN) {
+			ipoib_dbg_mcast(priv, "multicast join failed for %pI6, status %d\n",
+					mcast->mcmember.mgid.raw, status);
+		} else {
+			ipoib_warn(priv, "multicast join failed for %pI6, status %d\n",
+				   mcast->mcmember.mgid.raw, status);
+		}
+	}
+
+	mcast->backoff *= 2;
+	if (mcast->backoff > IPOIB_MAX_BACKOFF_SECONDS)
+		mcast->backoff = IPOIB_MAX_BACKOFF_SECONDS;
+
+	/* Clear the busy flag so we try again */
+	status = test_and_clear_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags);
+
+	mutex_lock(&mcast_mutex);
+	spin_lock_irq(&priv->lock);
+	if (test_bit(IPOIB_MCAST_RUN, &priv->flags))
+		queue_delayed_work(ipoib_workqueue, &priv->mcast_task,
+				   mcast->backoff * HZ);
+	spin_unlock_irq(&priv->lock);
+	mutex_unlock(&mcast_mutex);
+
+	return status;
+}
+
+static void ipoib_mcast_join(struct net_device *dev, struct ipoib_mcast *mcast,
+			     int create)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ib_sa_mcmember_rec rec = {
+		.join_state = 1
+	};
+	ib_sa_comp_mask comp_mask;
+	int ret = 0;
+
+	ipoib_dbg_mcast(priv, "joining MGID %pI6\n", mcast->mcmember.mgid.raw);
+
+	rec.mgid     = mcast->mcmember.mgid;
+	rec.port_gid = priv->local_gid;
+	rec.pkey     = cpu_to_be16(priv->pkey);
+
+	comp_mask =
+		IB_SA_MCMEMBER_REC_MGID		|
+		IB_SA_MCMEMBER_REC_PORT_GID	|
+		IB_SA_MCMEMBER_REC_PKEY		|
+		IB_SA_MCMEMBER_REC_JOIN_STATE;
+
+	if (create) {
+		comp_mask |=
+			IB_SA_MCMEMBER_REC_QKEY			|
+			IB_SA_MCMEMBER_REC_MTU_SELECTOR		|
+			IB_SA_MCMEMBER_REC_MTU			|
+			IB_SA_MCMEMBER_REC_TRAFFIC_CLASS	|
+			IB_SA_MCMEMBER_REC_RATE_SELECTOR	|
+			IB_SA_MCMEMBER_REC_RATE			|
+			IB_SA_MCMEMBER_REC_SL			|
+			IB_SA_MCMEMBER_REC_FLOW_LABEL		|
+			IB_SA_MCMEMBER_REC_HOP_LIMIT;
+
+		rec.qkey	  = priv->broadcast->mcmember.qkey;
+		rec.mtu_selector  = IB_SA_EQ;
+		rec.mtu		  = priv->broadcast->mcmember.mtu;
+		rec.traffic_class = priv->broadcast->mcmember.traffic_class;
+		rec.rate_selector = IB_SA_EQ;
+		rec.rate	  = priv->broadcast->mcmember.rate;
+		rec.sl		  = priv->broadcast->mcmember.sl;
+		rec.flow_label	  = priv->broadcast->mcmember.flow_label;
+		rec.hop_limit	  = priv->broadcast->mcmember.hop_limit;
+	}
+
+	set_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags);
+	mcast->mc = ib_sa_join_multicast(&ipoib_sa_client, priv->ca, priv->port,
+					 &rec, comp_mask, GFP_KERNEL,
+					 ipoib_mcast_join_complete, mcast);
+	if (IS_ERR(mcast->mc)) {
+		clear_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags);
+		ret = PTR_ERR(mcast->mc);
+		ipoib_warn(priv, "ib_sa_join_multicast failed, status %d\n", ret);
+
+		mcast->backoff *= 2;
+		if (mcast->backoff > IPOIB_MAX_BACKOFF_SECONDS)
+			mcast->backoff = IPOIB_MAX_BACKOFF_SECONDS;
+
+		mutex_lock(&mcast_mutex);
+		if (test_bit(IPOIB_MCAST_RUN, &priv->flags))
+			queue_delayed_work(ipoib_workqueue,
+					   &priv->mcast_task,
+					   mcast->backoff * HZ);
+		mutex_unlock(&mcast_mutex);
+	}
+}
+
+void ipoib_mcast_join_task(struct work_struct *work)
+{
+	struct ipoib_dev_priv *priv =
+		container_of(work, struct ipoib_dev_priv, mcast_task.work);
+	struct net_device *dev = priv->dev;
+
+	if (!test_bit(IPOIB_MCAST_RUN, &priv->flags))
+		return;
+
+	if (ib_query_gid(priv->ca, priv->port, 0, &priv->local_gid))
+		ipoib_warn(priv, "ib_query_gid() failed\n");
+	else
+		memcpy(priv->dev->dev_addr + 4, priv->local_gid.raw, sizeof (union ib_gid));
+
+	{
+		struct ib_port_attr attr;
+
+		if (!ib_query_port(priv->ca, priv->port, &attr))
+			priv->local_lid = attr.lid;
+		else
+			ipoib_warn(priv, "ib_query_port failed\n");
+	}
+
+	if (!priv->broadcast) {
+		struct ipoib_mcast *broadcast;
+
+		if (!test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
+			return;
+
+		broadcast = ipoib_mcast_alloc(dev, 1);
+		if (!broadcast) {
+			ipoib_warn(priv, "failed to allocate broadcast group\n");
+			mutex_lock(&mcast_mutex);
+			if (test_bit(IPOIB_MCAST_RUN, &priv->flags))
+				queue_delayed_work(ipoib_workqueue,
+						   &priv->mcast_task, HZ);
+			mutex_unlock(&mcast_mutex);
+			return;
+		}
+
+		spin_lock_irq(&priv->lock);
+		memcpy(broadcast->mcmember.mgid.raw, priv->dev->broadcast + 4,
+		       sizeof (union ib_gid));
+		priv->broadcast = broadcast;
+
+		__ipoib_mcast_add(dev, priv->broadcast);
+		spin_unlock_irq(&priv->lock);
+	}
+
+	if (!test_bit(IPOIB_MCAST_FLAG_ATTACHED, &priv->broadcast->flags)) {
+		if (!test_bit(IPOIB_MCAST_FLAG_BUSY, &priv->broadcast->flags))
+			ipoib_mcast_join(dev, priv->broadcast, 0);
+		return;
+	}
+
+	while (1) {
+		struct ipoib_mcast *mcast = NULL;
+
+		spin_lock_irq(&priv->lock);
+		list_for_each_entry(mcast, &priv->multicast_list, list) {
+			if (!test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags)
+			    && !test_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags)
+			    && !test_bit(IPOIB_MCAST_FLAG_ATTACHED, &mcast->flags)) {
+				/* Found the next unjoined group */
+				break;
+			}
+		}
+		spin_unlock_irq(&priv->lock);
+
+		if (&mcast->list == &priv->multicast_list) {
+			/* All done */
+			break;
+		}
+
+		ipoib_mcast_join(dev, mcast, 1);
+		return;
+	}
+
+	priv->mcast_mtu = IPOIB_UD_MTU(ib_mtu_enum_to_int(priv->broadcast->mcmember.mtu));
+
+	if (!ipoib_cm_admin_enabled(dev)) {
+		rtnl_lock();
+		dev_set_mtu(dev, min(priv->mcast_mtu, priv->admin_mtu));
+		rtnl_unlock();
+	}
+
+	ipoib_dbg_mcast(priv, "successfully joined all multicast groups\n");
+
+	clear_bit(IPOIB_MCAST_RUN, &priv->flags);
+}
+
+int ipoib_mcast_start_thread(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	ipoib_dbg_mcast(priv, "starting multicast thread\n");
+
+	mutex_lock(&mcast_mutex);
+	if (!test_and_set_bit(IPOIB_MCAST_RUN, &priv->flags))
+		queue_delayed_work(ipoib_workqueue, &priv->mcast_task, 0);
+	mutex_unlock(&mcast_mutex);
+
+	return 0;
+}
+
+int ipoib_mcast_stop_thread(struct net_device *dev, int flush)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	ipoib_dbg_mcast(priv, "stopping multicast thread\n");
+
+	mutex_lock(&mcast_mutex);
+	clear_bit(IPOIB_MCAST_RUN, &priv->flags);
+	cancel_delayed_work(&priv->mcast_task);
+	mutex_unlock(&mcast_mutex);
+
+	if (flush)
+		flush_workqueue(ipoib_workqueue);
+
+	return 0;
+}
+
+static int ipoib_mcast_leave(struct net_device *dev, struct ipoib_mcast *mcast)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	int ret = 0;
+
+	if (test_and_clear_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags))
+		ib_sa_free_multicast(mcast->mc);
+
+	if (test_and_clear_bit(IPOIB_MCAST_FLAG_ATTACHED, &mcast->flags)) {
+		ipoib_dbg_mcast(priv, "leaving MGID %pI6\n",
+				mcast->mcmember.mgid.raw);
+
+		/* Remove ourselves from the multicast group */
+		ret = ib_detach_mcast(priv->qp, &mcast->mcmember.mgid,
+				      be16_to_cpu(mcast->mcmember.mlid));
+		if (ret)
+			ipoib_warn(priv, "ib_detach_mcast failed (result = %d)\n", ret);
+	}
+
+	return 0;
+}
+
+void ipoib_mcast_send(struct net_device *dev, void *mgid, struct sk_buff *skb)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ipoib_mcast *mcast;
+	unsigned long flags;
+
+	spin_lock_irqsave(&priv->lock, flags);
+
+	if (!test_bit(IPOIB_FLAG_OPER_UP, &priv->flags)		||
+	    !priv->broadcast					||
+	    !test_bit(IPOIB_MCAST_FLAG_ATTACHED, &priv->broadcast->flags)) {
+		++dev->stats.tx_dropped;
+		dev_kfree_skb_any(skb);
+		goto unlock;
+	}
+
+	mcast = __ipoib_mcast_find(dev, mgid);
+	if (!mcast) {
+		/* Let's create a new send only group now */
+		ipoib_dbg_mcast(priv, "setting up send only multicast group for %pI6\n",
+				mgid);
+
+		mcast = ipoib_mcast_alloc(dev, 0);
+		if (!mcast) {
+			ipoib_warn(priv, "unable to allocate memory for "
+				   "multicast structure\n");
+			++dev->stats.tx_dropped;
+			dev_kfree_skb_any(skb);
+			goto out;
+		}
+
+		set_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags);
+		memcpy(mcast->mcmember.mgid.raw, mgid, sizeof (union ib_gid));
+		__ipoib_mcast_add(dev, mcast);
+		list_add_tail(&mcast->list, &priv->multicast_list);
+	}
+
+	if (!mcast->ah) {
+		if (skb_queue_len(&mcast->pkt_queue) < IPOIB_MAX_MCAST_QUEUE)
+			skb_queue_tail(&mcast->pkt_queue, skb);
+		else {
+			++dev->stats.tx_dropped;
+			dev_kfree_skb_any(skb);
+		}
+
+		if (test_bit(IPOIB_MCAST_FLAG_BUSY, &mcast->flags))
+			ipoib_dbg_mcast(priv, "no address vector, "
+					"but multicast join already started\n");
+		else if (test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags))
+			ipoib_mcast_sendonly_join(mcast);
+
+		/*
+		 * If lookup completes between here and out:, don't
+		 * want to send packet twice.
+		 */
+		mcast = NULL;
+	}
+
+out:
+	if (mcast && mcast->ah) {
+		if (skb_dst(skb)		&&
+		    skb_dst(skb)->neighbour &&
+		    !*to_ipoib_neigh(skb_dst(skb)->neighbour)) {
+			struct ipoib_neigh *neigh = ipoib_neigh_alloc(skb_dst(skb)->neighbour,
+									skb->dev);
+
+			if (neigh) {
+				kref_get(&mcast->ah->ref);
+				neigh->ah	= mcast->ah;
+				list_add_tail(&neigh->list, &mcast->neigh_list);
+			}
+		}
+
+		spin_unlock_irqrestore(&priv->lock, flags);
+		ipoib_send(dev, skb, mcast->ah, IB_MULTICAST_QPN);
+		return;
+	}
+
+unlock:
+	spin_unlock_irqrestore(&priv->lock, flags);
+}
+
+void ipoib_mcast_dev_flush(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	LIST_HEAD(remove_list);
+	struct ipoib_mcast *mcast, *tmcast;
+	unsigned long flags;
+
+	ipoib_dbg_mcast(priv, "flushing multicast list\n");
+
+	spin_lock_irqsave(&priv->lock, flags);
+
+	list_for_each_entry_safe(mcast, tmcast, &priv->multicast_list, list) {
+		list_del(&mcast->list);
+		rb_erase(&mcast->rb_node, &priv->multicast_tree);
+		list_add_tail(&mcast->list, &remove_list);
+	}
+
+	if (priv->broadcast) {
+		rb_erase(&priv->broadcast->rb_node, &priv->multicast_tree);
+		list_add_tail(&priv->broadcast->list, &remove_list);
+		priv->broadcast = NULL;
+	}
+
+	spin_unlock_irqrestore(&priv->lock, flags);
+
+	list_for_each_entry_safe(mcast, tmcast, &remove_list, list) {
+		ipoib_mcast_leave(dev, mcast);
+		ipoib_mcast_free(mcast);
+	}
+}
+
+static int ipoib_mcast_addr_is_valid(const u8 *addr, unsigned int addrlen,
+				     const u8 *broadcast)
+{
+	if (addrlen != INFINIBAND_ALEN)
+		return 0;
+	/* reserved QPN, prefix, scope */
+	if (memcmp(addr, broadcast, 6))
+		return 0;
+	/* signature lower, pkey */
+	if (memcmp(addr + 7, broadcast + 7, 3))
+		return 0;
+	return 1;
+}
+
+void ipoib_mcast_restart_task(struct work_struct *work)
+{
+	struct ipoib_dev_priv *priv =
+		container_of(work, struct ipoib_dev_priv, restart_task);
+	struct net_device *dev = priv->dev;
+	struct dev_mc_list *mclist;
+	struct ipoib_mcast *mcast, *tmcast;
+	LIST_HEAD(remove_list);
+	unsigned long flags;
+	struct ib_sa_mcmember_rec rec;
+
+	ipoib_dbg_mcast(priv, "restarting multicast task\n");
+
+	ipoib_mcast_stop_thread(dev, 0);
+
+	local_irq_save(flags);
+	netif_addr_lock(dev);
+	spin_lock(&priv->lock);
+
+	/*
+	 * Unfortunately, the networking core only gives us a list of all of
+	 * the multicast hardware addresses. We need to figure out which ones
+	 * are new and which ones have been removed
+	 */
+
+	/* Clear out the found flag */
+	list_for_each_entry(mcast, &priv->multicast_list, list)
+		clear_bit(IPOIB_MCAST_FLAG_FOUND, &mcast->flags);
+
+	/* Mark all of the entries that are found or don't exist */
+	for (mclist = dev->mc_list; mclist; mclist = mclist->next) {
+		union ib_gid mgid;
+
+		if (!ipoib_mcast_addr_is_valid(mclist->dmi_addr,
+					       mclist->dmi_addrlen,
+					       dev->broadcast))
+			continue;
+
+		memcpy(mgid.raw, mclist->dmi_addr + 4, sizeof mgid);
+
+		mcast = __ipoib_mcast_find(dev, &mgid);
+		if (!mcast || test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags)) {
+			struct ipoib_mcast *nmcast;
+
+			/* ignore group which is directly joined by userspace */
+			if (test_bit(IPOIB_FLAG_UMCAST, &priv->flags) &&
+			    !ib_sa_get_mcmember_rec(priv->ca, priv->port, &mgid, &rec)) {
+				ipoib_dbg_mcast(priv, "ignoring multicast entry for mgid %pI6\n",
+						mgid.raw);
+				continue;
+			}
+
+			/* Not found or send-only group, let's add a new entry */
+			ipoib_dbg_mcast(priv, "adding multicast entry for mgid %pI6\n",
+					mgid.raw);
+
+			nmcast = ipoib_mcast_alloc(dev, 0);
+			if (!nmcast) {
+				ipoib_warn(priv, "unable to allocate memory for multicast structure\n");
+				continue;
+			}
+
+			set_bit(IPOIB_MCAST_FLAG_FOUND, &nmcast->flags);
+
+			nmcast->mcmember.mgid = mgid;
+
+			if (mcast) {
+				/* Destroy the send only entry */
+				list_move_tail(&mcast->list, &remove_list);
+
+				rb_replace_node(&mcast->rb_node,
+						&nmcast->rb_node,
+						&priv->multicast_tree);
+			} else
+				__ipoib_mcast_add(dev, nmcast);
+
+			list_add_tail(&nmcast->list, &priv->multicast_list);
+		}
+
+		if (mcast)
+			set_bit(IPOIB_MCAST_FLAG_FOUND, &mcast->flags);
+	}
+
+	/* Remove all of the entries don't exist anymore */
+	list_for_each_entry_safe(mcast, tmcast, &priv->multicast_list, list) {
+		if (!test_bit(IPOIB_MCAST_FLAG_FOUND, &mcast->flags) &&
+		    !test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags)) {
+			ipoib_dbg_mcast(priv, "deleting multicast group %pI6\n",
+					mcast->mcmember.mgid.raw);
+
+			rb_erase(&mcast->rb_node, &priv->multicast_tree);
+
+			/* Move to the remove list */
+			list_move_tail(&mcast->list, &remove_list);
+		}
+	}
+
+	spin_unlock(&priv->lock);
+	netif_addr_unlock(dev);
+	local_irq_restore(flags);
+
+	/* We have to cancel outside of the spinlock */
+	list_for_each_entry_safe(mcast, tmcast, &remove_list, list) {
+		ipoib_mcast_leave(mcast->dev, mcast);
+		ipoib_mcast_free(mcast);
+	}
+
+	if (test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
+		ipoib_mcast_start_thread(dev);
+}
+
+#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
+
+struct ipoib_mcast_iter *ipoib_mcast_iter_init(struct net_device *dev)
+{
+	struct ipoib_mcast_iter *iter;
+
+	iter = kmalloc(sizeof *iter, GFP_KERNEL);
+	if (!iter)
+		return NULL;
+
+	iter->dev = dev;
+	memset(iter->mgid.raw, 0, 16);
+
+	if (ipoib_mcast_iter_next(iter)) {
+		kfree(iter);
+		return NULL;
+	}
+
+	return iter;
+}
+
+int ipoib_mcast_iter_next(struct ipoib_mcast_iter *iter)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(iter->dev);
+	struct rb_node *n;
+	struct ipoib_mcast *mcast;
+	int ret = 1;
+
+	spin_lock_irq(&priv->lock);
+
+	n = rb_first(&priv->multicast_tree);
+
+	while (n) {
+		mcast = rb_entry(n, struct ipoib_mcast, rb_node);
+
+		if (memcmp(iter->mgid.raw, mcast->mcmember.mgid.raw,
+			   sizeof (union ib_gid)) < 0) {
+			iter->mgid      = mcast->mcmember.mgid;
+			iter->created   = mcast->created;
+			iter->queuelen  = skb_queue_len(&mcast->pkt_queue);
+			iter->complete  = !!mcast->ah;
+			iter->send_only = !!(mcast->flags & (1 << IPOIB_MCAST_FLAG_SENDONLY));
+
+			ret = 0;
+
+			break;
+		}
+
+		n = rb_next(n);
+	}
+
+	spin_unlock_irq(&priv->lock);
+
+	return ret;
+}
+
+void ipoib_mcast_iter_read(struct ipoib_mcast_iter *iter,
+			   union ib_gid *mgid,
+			   unsigned long *created,
+			   unsigned int *queuelen,
+			   unsigned int *complete,
+			   unsigned int *send_only)
+{
+	*mgid      = iter->mgid;
+	*created   = iter->created;
+	*queuelen  = iter->queuelen;
+	*complete  = iter->complete;
+	*send_only = iter->send_only;
+}
+
+#endif /* CONFIG_INFINIBAND_IPOIB_DEBUG */

kernel/drivers/infiniband/ulp/ipoib/ipoib_verbs.c

@@ -0,0 +1,292 @@
+/*
+ * Copyright (c) 2004, 2005 Topspin Communications.  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.
+ */
+
+#include "ipoib.h"
+
+int ipoib_mcast_attach(struct net_device *dev, u16 mlid, union ib_gid *mgid, int set_qkey)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ib_qp_attr *qp_attr = NULL;
+	int ret;
+	u16 pkey_index;
+
+	if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index)) {
+		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
+		ret = -ENXIO;
+		goto out;
+	}
+	set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
+
+	if (set_qkey) {
+		ret = -ENOMEM;
+		qp_attr = kmalloc(sizeof *qp_attr, GFP_KERNEL);
+		if (!qp_attr)
+			goto out;
+
+		/* set correct QKey for QP */
+		qp_attr->qkey = priv->qkey;
+		ret = ib_modify_qp(priv->qp, qp_attr, IB_QP_QKEY);
+		if (ret) {
+			ipoib_warn(priv, "failed to modify QP, ret = %d\n", ret);
+			goto out;
+		}
+	}
+
+	/* attach QP to multicast group */
+	ret = ib_attach_mcast(priv->qp, mgid, mlid);
+	if (ret)
+		ipoib_warn(priv, "failed to attach to multicast group, ret = %d\n", ret);
+
+out:
+	kfree(qp_attr);
+	return ret;
+}
+
+int ipoib_init_qp(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	int ret;
+	struct ib_qp_attr qp_attr;
+	int attr_mask;
+
+	if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags))
+		return -1;
+
+	qp_attr.qp_state = IB_QPS_INIT;
+	qp_attr.qkey = 0;
+	qp_attr.port_num = priv->port;
+	qp_attr.pkey_index = priv->pkey_index;
+	attr_mask =
+	    IB_QP_QKEY |
+	    IB_QP_PORT |
+	    IB_QP_PKEY_INDEX |
+	    IB_QP_STATE;
+	ret = ib_modify_qp(priv->qp, &qp_attr, attr_mask);
+	if (ret) {
+		ipoib_warn(priv, "failed to modify QP to init, ret = %d\n", ret);
+		goto out_fail;
+	}
+
+	qp_attr.qp_state = IB_QPS_RTR;
+	/* Can't set this in a INIT->RTR transition */
+	attr_mask &= ~IB_QP_PORT;
+	ret = ib_modify_qp(priv->qp, &qp_attr, attr_mask);
+	if (ret) {
+		ipoib_warn(priv, "failed to modify QP to RTR, ret = %d\n", ret);
+		goto out_fail;
+	}
+
+	qp_attr.qp_state = IB_QPS_RTS;
+	qp_attr.sq_psn = 0;
+	attr_mask |= IB_QP_SQ_PSN;
+	attr_mask &= ~IB_QP_PKEY_INDEX;
+	ret = ib_modify_qp(priv->qp, &qp_attr, attr_mask);
+	if (ret) {
+		ipoib_warn(priv, "failed to modify QP to RTS, ret = %d\n", ret);
+		goto out_fail;
+	}
+
+	return 0;
+
+out_fail:
+	qp_attr.qp_state = IB_QPS_RESET;
+	if (ib_modify_qp(priv->qp, &qp_attr, IB_QP_STATE))
+		ipoib_warn(priv, "Failed to modify QP to RESET state\n");
+
+	return ret;
+}
+
+int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+	struct ib_qp_init_attr init_attr = {
+		.cap = {
+			.max_send_wr  = ipoib_sendq_size,
+			.max_recv_wr  = ipoib_recvq_size,
+			.max_send_sge = 1,
+			.max_recv_sge = IPOIB_UD_RX_SG
+		},
+		.sq_sig_type = IB_SIGNAL_ALL_WR,
+		.qp_type     = IB_QPT_UD
+	};
+
+	int ret, size;
+	int i;
+
+	priv->pd = ib_alloc_pd(priv->ca);
+	if (IS_ERR(priv->pd)) {
+		printk(KERN_WARNING "%s: failed to allocate PD\n", ca->name);
+		return -ENODEV;
+	}
+
+	priv->mr = ib_get_dma_mr(priv->pd, IB_ACCESS_LOCAL_WRITE);
+	if (IS_ERR(priv->mr)) {
+		printk(KERN_WARNING "%s: ib_get_dma_mr failed\n", ca->name);
+		goto out_free_pd;
+	}
+
+	size = ipoib_recvq_size + 1;
+	ret = ipoib_cm_dev_init(dev);
+	if (!ret) {
+		size += ipoib_sendq_size;
+		if (ipoib_cm_has_srq(dev))
+			size += ipoib_recvq_size + 1; /* 1 extra for rx_drain_qp */
+		else
+			size += ipoib_recvq_size * ipoib_max_conn_qp;
+	}
+
+	priv->recv_cq = ib_create_cq(priv->ca, ipoib_ib_completion, NULL, dev, size, 0);
+	if (IS_ERR(priv->recv_cq)) {
+		printk(KERN_WARNING "%s: failed to create receive CQ\n", ca->name);
+		goto out_free_mr;
+	}
+
+	priv->send_cq = ib_create_cq(priv->ca, ipoib_send_comp_handler, NULL,
+				     dev, ipoib_sendq_size, 0);
+	if (IS_ERR(priv->send_cq)) {
+		printk(KERN_WARNING "%s: failed to create send CQ\n", ca->name);
+		goto out_free_recv_cq;
+	}
+
+	if (ib_req_notify_cq(priv->recv_cq, IB_CQ_NEXT_COMP))
+		goto out_free_send_cq;
+
+	init_attr.send_cq = priv->send_cq;
+	init_attr.recv_cq = priv->recv_cq;
+
+	if (priv->hca_caps & IB_DEVICE_UD_TSO)
+		init_attr.create_flags |= IB_QP_CREATE_IPOIB_UD_LSO;
+
+	if (priv->hca_caps & IB_DEVICE_BLOCK_MULTICAST_LOOPBACK)
+		init_attr.create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK;
+
+	if (dev->features & NETIF_F_SG)
+		init_attr.cap.max_send_sge = MAX_SKB_FRAGS + 1;
+
+	priv->qp = ib_create_qp(priv->pd, &init_attr);
+	if (IS_ERR(priv->qp)) {
+		printk(KERN_WARNING "%s: failed to create QP\n", ca->name);
+		goto out_free_send_cq;
+	}
+
+	priv->dev->dev_addr[1] = (priv->qp->qp_num >> 16) & 0xff;
+	priv->dev->dev_addr[2] = (priv->qp->qp_num >>  8) & 0xff;
+	priv->dev->dev_addr[3] = (priv->qp->qp_num      ) & 0xff;
+
+	for (i = 0; i < MAX_SKB_FRAGS + 1; ++i)
+		priv->tx_sge[i].lkey = priv->mr->lkey;
+
+	priv->tx_wr.opcode	= IB_WR_SEND;
+	priv->tx_wr.sg_list	= priv->tx_sge;
+	priv->tx_wr.send_flags	= IB_SEND_SIGNALED;
+
+	priv->rx_sge[0].lkey = priv->mr->lkey;
+	if (ipoib_ud_need_sg(priv->max_ib_mtu)) {
+		priv->rx_sge[0].length = IPOIB_UD_HEAD_SIZE;
+		priv->rx_sge[1].length = PAGE_SIZE;
+		priv->rx_sge[1].lkey = priv->mr->lkey;
+		priv->rx_wr.num_sge = IPOIB_UD_RX_SG;
+	} else {
+		priv->rx_sge[0].length = IPOIB_UD_BUF_SIZE(priv->max_ib_mtu);
+		priv->rx_wr.num_sge = 1;
+	}
+	priv->rx_wr.next = NULL;
+	priv->rx_wr.sg_list = priv->rx_sge;
+
+	return 0;
+
+out_free_send_cq:
+	ib_destroy_cq(priv->send_cq);
+
+out_free_recv_cq:
+	ib_destroy_cq(priv->recv_cq);
+
+out_free_mr:
+	ib_dereg_mr(priv->mr);
+	ipoib_cm_dev_cleanup(dev);
+
+out_free_pd:
+	ib_dealloc_pd(priv->pd);
+	return -ENODEV;
+}
+
+void ipoib_transport_dev_cleanup(struct net_device *dev)
+{
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	if (priv->qp) {
+		if (ib_destroy_qp(priv->qp))
+			ipoib_warn(priv, "ib_qp_destroy failed\n");
+
+		priv->qp = NULL;
+		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
+	}
+
+	if (ib_destroy_cq(priv->send_cq))
+		ipoib_warn(priv, "ib_cq_destroy (send) failed\n");
+
+	if (ib_destroy_cq(priv->recv_cq))
+		ipoib_warn(priv, "ib_cq_destroy (recv) failed\n");
+
+	ipoib_cm_dev_cleanup(dev);
+
+	if (ib_dereg_mr(priv->mr))
+		ipoib_warn(priv, "ib_dereg_mr failed\n");
+
+	if (ib_dealloc_pd(priv->pd))
+		ipoib_warn(priv, "ib_dealloc_pd failed\n");
+}
+
+void ipoib_event(struct ib_event_handler *handler,
+		 struct ib_event *record)
+{
+	struct ipoib_dev_priv *priv =
+		container_of(handler, struct ipoib_dev_priv, event_handler);
+
+	if (record->element.port_num != priv->port)
+		return;
+
+	ipoib_dbg(priv, "Event %d on device %s port %d\n", record->event,
+		  record->device->name, record->element.port_num);
+
+	if (record->event == IB_EVENT_SM_CHANGE ||
+	    record->event == IB_EVENT_CLIENT_REREGISTER) {
+		queue_work(ipoib_workqueue, &priv->flush_light);
+	} else if (record->event == IB_EVENT_PORT_ERR ||
+		   record->event == IB_EVENT_PORT_ACTIVE ||
+		   record->event == IB_EVENT_LID_CHANGE) {
+		queue_work(ipoib_workqueue, &priv->flush_normal);
+	} else if (record->event == IB_EVENT_PKEY_CHANGE) {
+		queue_work(ipoib_workqueue, &priv->flush_heavy);
+	}
+}

kernel/drivers/infiniband/ulp/ipoib/ipoib_vlan.c

@@ -0,0 +1,192 @@
+/*
+ * 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.
+ */
+
+#include <linux/module.h>
+
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/seq_file.h>
+
+#include <asm/uaccess.h>
+
+#include "ipoib.h"
+
+static ssize_t show_parent(struct device *d, struct device_attribute *attr,
+			   char *buf)
+{
+	struct net_device *dev = to_net_dev(d);
+	struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+	return sprintf(buf, "%s\n", priv->parent->name);
+}
+static DEVICE_ATTR(parent, S_IRUGO, show_parent, NULL);
+
+int ipoib_vlan_add(struct net_device *pdev, unsigned short pkey)
+{
+	struct ipoib_dev_priv *ppriv, *priv;
+	char intf_name[IFNAMSIZ];
+	int result;
+
+	if (!capable(CAP_NET_ADMIN))
+		return -EPERM;
+
+	ppriv = netdev_priv(pdev);
+
+	if (!rtnl_trylock())
+		return restart_syscall();
+	mutex_lock(&ppriv->vlan_mutex);
+
+	/*
+	 * First ensure this isn't a duplicate. We check the parent device and
+	 * then all of the child interfaces to make sure the Pkey doesn't match.
+	 */
+	if (ppriv->pkey == pkey) {
+		result = -ENOTUNIQ;
+		priv = NULL;
+		goto err;
+	}
+
+	list_for_each_entry(priv, &ppriv->child_intfs, list) {
+		if (priv->pkey == pkey) {
+			result = -ENOTUNIQ;
+			priv = NULL;
+			goto err;
+		}
+	}
+
+	snprintf(intf_name, sizeof intf_name, "%s.%04x",
+		 ppriv->dev->name, pkey);
+	priv = ipoib_intf_alloc(intf_name);
+	if (!priv) {
+		result = -ENOMEM;
+		goto err;
+	}
+
+	priv->max_ib_mtu = ppriv->max_ib_mtu;
+	/* MTU will be reset when mcast join happens */
+	priv->dev->mtu   = IPOIB_UD_MTU(priv->max_ib_mtu);
+	priv->mcast_mtu  = priv->admin_mtu = priv->dev->mtu;
+	set_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags);
+
+	result = ipoib_set_dev_features(priv, ppriv->ca);
+	if (result)
+		goto err;
+
+	priv->pkey = pkey;
+
+	memcpy(priv->dev->dev_addr, ppriv->dev->dev_addr, INFINIBAND_ALEN);
+	priv->dev->broadcast[8] = pkey >> 8;
+	priv->dev->broadcast[9] = pkey & 0xff;
+
+	result = ipoib_dev_init(priv->dev, ppriv->ca, ppriv->port);
+	if (result < 0) {
+		ipoib_warn(ppriv, "failed to initialize subinterface: "
+			   "device %s, port %d",
+			   ppriv->ca->name, ppriv->port);
+		goto err;
+	}
+
+	result = register_netdevice(priv->dev);
+	if (result) {
+		ipoib_warn(priv, "failed to initialize; error %i", result);
+		goto register_failed;
+	}
+
+	priv->parent = ppriv->dev;
+
+	ipoib_create_debug_files(priv->dev);
+
+	if (ipoib_cm_add_mode_attr(priv->dev))
+		goto sysfs_failed;
+	if (ipoib_add_pkey_attr(priv->dev))
+		goto sysfs_failed;
+	if (ipoib_add_umcast_attr(priv->dev))
+		goto sysfs_failed;
+
+	if (device_create_file(&priv->dev->dev, &dev_attr_parent))
+		goto sysfs_failed;
+
+	list_add_tail(&priv->list, &ppriv->child_intfs);
+
+	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
+
+	return 0;
+
+sysfs_failed:
+	ipoib_delete_debug_files(priv->dev);
+	unregister_netdevice(priv->dev);
+
+register_failed:
+	ipoib_dev_cleanup(priv->dev);
+
+err:
+	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
+	if (priv)
+		free_netdev(priv->dev);
+
+	return result;
+}
+
+int ipoib_vlan_delete(struct net_device *pdev, unsigned short pkey)
+{
+	struct ipoib_dev_priv *ppriv, *priv, *tpriv;
+	struct net_device *dev = NULL;
+
+	if (!capable(CAP_NET_ADMIN))
+		return -EPERM;
+
+	ppriv = netdev_priv(pdev);
+
+	if (!rtnl_trylock())
+		return restart_syscall();
+	mutex_lock(&ppriv->vlan_mutex);
+	list_for_each_entry_safe(priv, tpriv, &ppriv->child_intfs, list) {
+		if (priv->pkey == pkey) {
+			unregister_netdevice(priv->dev);
+			ipoib_dev_cleanup(priv->dev);
+			list_del(&priv->list);
+			dev = priv->dev;
+			break;
+		}
+	}
+	mutex_unlock(&ppriv->vlan_mutex);
+	rtnl_unlock();
+
+	if (dev) {
+		free_netdev(dev);
+		return 0;
+	}
+
+	return -ENODEV;
+}

kernel/drivers/infiniband/ulp/iser/Kconfig

@@ -0,0 +1,12 @@
+config INFINIBAND_ISER
+	tristate "iSCSI Extensions for RDMA (iSER)"
+	depends on SCSI && INET && INFINIBAND_ADDR_TRANS
+	select SCSI_ISCSI_ATTRS
+	---help---
+	  Support for the iSCSI Extensions for RDMA (iSER) Protocol
+          over InfiniBand. This allows you to access storage devices
+          that speak iSCSI over iSER over InfiniBand.
+
+	  The iSER protocol is defined by IETF.
+	  See <http://www.ietf.org/rfc/rfc5046.txt>
+	  and <http://www.infinibandta.org/members/spec/Annex_iSER.PDF>

kernel/drivers/infiniband/ulp/iser/Makefile

@@ -0,0 +1,4 @@
+obj-$(CONFIG_INFINIBAND_ISER)	+= ib_iser.o
+
+ib_iser-y			:= iser_verbs.o iser_initiator.o iser_memory.o \
+				   iscsi_iser.o

kernel/drivers/infiniband/ulp/iser/iscsi_iser.c

@@ -0,0 +1,713 @@
+/*
+ * iSCSI Initiator over iSER Data-Path
+ *
+ * Copyright (C) 2004 Dmitry Yusupov
+ * Copyright (C) 2004 Alex Aizman
+ * Copyright (C) 2005 Mike Christie
+ * Copyright (c) 2005, 2006 Voltaire, Inc. All rights reserved.
+ * maintained by openib-general@openib.org
+ *
+ * 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.
+ *
+ * Credits:
+ *	Christoph Hellwig
+ *	FUJITA Tomonori
+ *	Arne Redlich
+ *	Zhenyu Wang
+ * Modified by:
+ *      Erez Zilber
+ */
+
+#include <linux/types.h>
+#include <linux/list.h>
+#include <linux/hardirq.h>
+#include <linux/kfifo.h>
+#include <linux/blkdev.h>
+#include <linux/init.h>
+#include <linux/ioctl.h>
+#include <linux/cdev.h>
+#include <linux/in.h>
+#include <linux/net.h>
+#include <linux/scatterlist.h>
+#include <linux/delay.h>
+
+#include <net/sock.h>
+
+#include <asm/uaccess.h>
+
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_eh.h>
+#include <scsi/scsi_tcq.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_transport_iscsi.h>
+
+#include "iscsi_iser.h"
+
+static struct scsi_host_template iscsi_iser_sht;
+static struct iscsi_transport iscsi_iser_transport;
+static struct scsi_transport_template *iscsi_iser_scsi_transport;
+
+static unsigned int iscsi_max_lun = 512;
+module_param_named(max_lun, iscsi_max_lun, uint, S_IRUGO);
+
+int iser_debug_level = 0;
+
+MODULE_DESCRIPTION("iSER (iSCSI Extensions for RDMA) Datamover "
+		   "v" DRV_VER " (" DRV_DATE ")");
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_AUTHOR("Alex Nezhinsky, Dan Bar Dov, Or Gerlitz");
+
+module_param_named(debug_level, iser_debug_level, int, 0644);
+MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0 (default:disabled)");
+
+struct iser_global ig;
+
+void
+iscsi_iser_recv(struct iscsi_conn *conn,
+		struct iscsi_hdr *hdr, char *rx_data, int rx_data_len)
+{
+	int rc = 0;
+	int datalen;
+	int ahslen;
+
+	/* verify PDU length */
+	datalen = ntoh24(hdr->dlength);
+	if (datalen != rx_data_len) {
+		printk(KERN_ERR "iscsi_iser: datalen %d (hdr) != %d (IB) \n",
+		       datalen, rx_data_len);
+		rc = ISCSI_ERR_DATALEN;
+		goto error;
+	}
+
+	/* read AHS */
+	ahslen = hdr->hlength * 4;
+
+	rc = iscsi_complete_pdu(conn, hdr, rx_data, rx_data_len);
+	if (rc && rc != ISCSI_ERR_NO_SCSI_CMD)
+		goto error;
+
+	return;
+error:
+	iscsi_conn_failure(conn, rc);
+}
+
+static int iscsi_iser_pdu_alloc(struct iscsi_task *task, uint8_t opcode)
+{
+	struct iscsi_iser_task *iser_task = task->dd_data;
+
+	task->hdr = (struct iscsi_hdr *)&iser_task->desc.iscsi_header;
+	task->hdr_max = sizeof(iser_task->desc.iscsi_header);
+	return 0;
+}
+
+/**
+ * iscsi_iser_task_init - Initialize task
+ * @task: iscsi task
+ *
+ * Initialize the task for the scsi command or mgmt command.
+ */
+static int
+iscsi_iser_task_init(struct iscsi_task *task)
+{
+	struct iscsi_iser_conn *iser_conn  = task->conn->dd_data;
+	struct iscsi_iser_task *iser_task = task->dd_data;
+
+	/* mgmt task */
+	if (!task->sc) {
+		iser_task->desc.data = task->data;
+		return 0;
+	}
+
+	iser_task->command_sent = 0;
+	iser_task->iser_conn    = iser_conn;
+	iser_task_rdma_init(iser_task);
+	return 0;
+}
+
+/**
+ * iscsi_iser_mtask_xmit - xmit management(immediate) task
+ * @conn: iscsi connection
+ * @task: task management task
+ *
+ * Notes:
+ *	The function can return -EAGAIN in which case caller must
+ *	call it again later, or recover. '0' return code means successful
+ *	xmit.
+ *
+ **/
+static int
+iscsi_iser_mtask_xmit(struct iscsi_conn *conn, struct iscsi_task *task)
+{
+	int error = 0;
+
+	iser_dbg("task deq [cid %d itt 0x%x]\n", conn->id, task->itt);
+
+	error = iser_send_control(conn, task);
+
+	/* since iser xmits control with zero copy, tasks can not be recycled
+	 * right after sending them.
+	 * The recycling scheme is based on whether a response is expected
+	 * - if yes, the task is recycled at iscsi_complete_pdu
+	 * - if no,  the task is recycled at iser_snd_completion
+	 */
+	if (error && error != -ENOBUFS)
+		iscsi_conn_failure(conn, ISCSI_ERR_CONN_FAILED);
+
+	return error;
+}
+
+static int
+iscsi_iser_task_xmit_unsol_data(struct iscsi_conn *conn,
+				 struct iscsi_task *task)
+{
+	struct iscsi_r2t_info *r2t = &task->unsol_r2t;
+	struct iscsi_data hdr;
+	int error = 0;
+
+	/* Send data-out PDUs while there's still unsolicited data to send */
+	while (iscsi_task_has_unsol_data(task)) {
+		iscsi_prep_data_out_pdu(task, r2t, &hdr);
+		iser_dbg("Sending data-out: itt 0x%x, data count %d\n",
+			   hdr.itt, r2t->data_count);
+
+		/* the buffer description has been passed with the command */
+		/* Send the command */
+		error = iser_send_data_out(conn, task, &hdr);
+		if (error) {
+			r2t->datasn--;
+			goto iscsi_iser_task_xmit_unsol_data_exit;
+		}
+		r2t->sent += r2t->data_count;
+		iser_dbg("Need to send %d more as data-out PDUs\n",
+			   r2t->data_length - r2t->sent);
+	}
+
+iscsi_iser_task_xmit_unsol_data_exit:
+	return error;
+}
+
+static int
+iscsi_iser_task_xmit(struct iscsi_task *task)
+{
+	struct iscsi_conn *conn = task->conn;
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	int error = 0;
+
+	if (!task->sc)
+		return iscsi_iser_mtask_xmit(conn, task);
+
+	if (task->sc->sc_data_direction == DMA_TO_DEVICE) {
+		BUG_ON(scsi_bufflen(task->sc) == 0);
+
+		iser_dbg("cmd [itt %x total %d imm %d unsol_data %d\n",
+			   task->itt, scsi_bufflen(task->sc),
+			   task->imm_count, task->unsol_r2t.data_length);
+	}
+
+	iser_dbg("task deq [cid %d itt 0x%x]\n",
+		   conn->id, task->itt);
+
+	/* Send the cmd PDU */
+	if (!iser_task->command_sent) {
+		error = iser_send_command(conn, task);
+		if (error)
+			goto iscsi_iser_task_xmit_exit;
+		iser_task->command_sent = 1;
+	}
+
+	/* Send unsolicited data-out PDU(s) if necessary */
+	if (iscsi_task_has_unsol_data(task))
+		error = iscsi_iser_task_xmit_unsol_data(conn, task);
+
+ iscsi_iser_task_xmit_exit:
+	if (error && error != -ENOBUFS)
+		iscsi_conn_failure(conn, ISCSI_ERR_CONN_FAILED);
+	return error;
+}
+
+static void iscsi_iser_cleanup_task(struct iscsi_task *task)
+{
+	struct iscsi_iser_task *iser_task = task->dd_data;
+
+	/* mgmt tasks do not need special cleanup */
+	if (!task->sc)
+		return;
+
+	if (iser_task->status == ISER_TASK_STATUS_STARTED) {
+		iser_task->status = ISER_TASK_STATUS_COMPLETED;
+		iser_task_rdma_finalize(iser_task);
+	}
+}
+
+static struct iscsi_cls_conn *
+iscsi_iser_conn_create(struct iscsi_cls_session *cls_session, uint32_t conn_idx)
+{
+	struct iscsi_conn *conn;
+	struct iscsi_cls_conn *cls_conn;
+	struct iscsi_iser_conn *iser_conn;
+
+	cls_conn = iscsi_conn_setup(cls_session, sizeof(*iser_conn), conn_idx);
+	if (!cls_conn)
+		return NULL;
+	conn = cls_conn->dd_data;
+
+	/*
+	 * due to issues with the login code re iser sematics
+	 * this not set in iscsi_conn_setup - FIXME
+	 */
+	conn->max_recv_dlength = 128;
+
+	iser_conn = conn->dd_data;
+	conn->dd_data = iser_conn;
+	iser_conn->iscsi_conn = conn;
+
+	return cls_conn;
+}
+
+static void
+iscsi_iser_conn_destroy(struct iscsi_cls_conn *cls_conn)
+{
+	struct iscsi_conn *conn = cls_conn->dd_data;
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iser_conn *ib_conn = iser_conn->ib_conn;
+
+	iscsi_conn_teardown(cls_conn);
+	/*
+	 * Userspace will normally call the stop callback and
+	 * already have freed the ib_conn, but if it goofed up then
+	 * we free it here.
+	 */
+	if (ib_conn) {
+		ib_conn->iser_conn = NULL;
+		iser_conn_put(ib_conn);
+	}
+}
+
+static int
+iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session,
+		     struct iscsi_cls_conn *cls_conn, uint64_t transport_eph,
+		     int is_leading)
+{
+	struct iscsi_conn *conn = cls_conn->dd_data;
+	struct iscsi_iser_conn *iser_conn;
+	struct iser_conn *ib_conn;
+	struct iscsi_endpoint *ep;
+	int error;
+
+	error = iscsi_conn_bind(cls_session, cls_conn, is_leading);
+	if (error)
+		return error;
+
+	/* the transport ep handle comes from user space so it must be
+	 * verified against the global ib connections list */
+	ep = iscsi_lookup_endpoint(transport_eph);
+	if (!ep) {
+		iser_err("can't bind eph %llx\n",
+			 (unsigned long long)transport_eph);
+		return -EINVAL;
+	}
+	ib_conn = ep->dd_data;
+
+	/* binds the iSER connection retrieved from the previously
+	 * connected ep_handle to the iSCSI layer connection. exchanges
+	 * connection pointers */
+	iser_err("binding iscsi conn %p to iser_conn %p\n",conn,ib_conn);
+	iser_conn = conn->dd_data;
+	ib_conn->iser_conn = iser_conn;
+	iser_conn->ib_conn  = ib_conn;
+	iser_conn_get(ib_conn);
+	return 0;
+}
+
+static void
+iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag)
+{
+	struct iscsi_conn *conn = cls_conn->dd_data;
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iser_conn *ib_conn = iser_conn->ib_conn;
+
+	/*
+	 * Userspace may have goofed up and not bound the connection or
+	 * might have only partially setup the connection.
+	 */
+	if (ib_conn) {
+		iscsi_conn_stop(cls_conn, flag);
+		/*
+		 * There is no unbind event so the stop callback
+		 * must release the ref from the bind.
+		 */
+		iser_conn_put(ib_conn);
+	}
+	iser_conn->ib_conn = NULL;
+}
+
+static int
+iscsi_iser_conn_start(struct iscsi_cls_conn *cls_conn)
+{
+	struct iscsi_conn *conn = cls_conn->dd_data;
+	int err;
+
+	err = iser_conn_set_full_featured_mode(conn);
+	if (err)
+		return err;
+
+	return iscsi_conn_start(cls_conn);
+}
+
+static void iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session)
+{
+	struct Scsi_Host *shost = iscsi_session_to_shost(cls_session);
+
+	iscsi_session_teardown(cls_session);
+	iscsi_host_remove(shost);
+	iscsi_host_free(shost);
+}
+
+static struct iscsi_cls_session *
+iscsi_iser_session_create(struct iscsi_endpoint *ep,
+			  uint16_t cmds_max, uint16_t qdepth,
+			  uint32_t initial_cmdsn)
+{
+	struct iscsi_cls_session *cls_session;
+	struct iscsi_session *session;
+	struct Scsi_Host *shost;
+	struct iser_conn *ib_conn;
+
+	shost = iscsi_host_alloc(&iscsi_iser_sht, 0, 1);
+	if (!shost)
+		return NULL;
+	shost->transportt = iscsi_iser_scsi_transport;
+	shost->max_lun = iscsi_max_lun;
+	shost->max_id = 0;
+	shost->max_channel = 0;
+	shost->max_cmd_len = 16;
+
+	/*
+	 * older userspace tools (before 2.0-870) did not pass us
+	 * the leading conn's ep so this will be NULL;
+	 */
+	if (ep)
+		ib_conn = ep->dd_data;
+
+	if (iscsi_host_add(shost,
+			   ep ? ib_conn->device->ib_device->dma_device : NULL))
+		goto free_host;
+
+	/*
+	 * we do not support setting can_queue cmd_per_lun from userspace yet
+	 * because we preallocate so many resources
+	 */
+	cls_session = iscsi_session_setup(&iscsi_iser_transport, shost,
+					  ISCSI_DEF_XMIT_CMDS_MAX, 0,
+					  sizeof(struct iscsi_iser_task),
+					  initial_cmdsn, 0);
+	if (!cls_session)
+		goto remove_host;
+	session = cls_session->dd_data;
+
+	shost->can_queue = session->scsi_cmds_max;
+	return cls_session;
+
+remove_host:
+	iscsi_host_remove(shost);
+free_host:
+	iscsi_host_free(shost);
+	return NULL;
+}
+
+static int
+iscsi_iser_set_param(struct iscsi_cls_conn *cls_conn,
+		     enum iscsi_param param, char *buf, int buflen)
+{
+	int value;
+
+	switch (param) {
+	case ISCSI_PARAM_MAX_RECV_DLENGTH:
+		/* TBD */
+		break;
+	case ISCSI_PARAM_HDRDGST_EN:
+		sscanf(buf, "%d", &value);
+		if (value) {
+			printk(KERN_ERR "DataDigest wasn't negotiated to None");
+			return -EPROTO;
+		}
+		break;
+	case ISCSI_PARAM_DATADGST_EN:
+		sscanf(buf, "%d", &value);
+		if (value) {
+			printk(KERN_ERR "DataDigest wasn't negotiated to None");
+			return -EPROTO;
+		}
+		break;
+	case ISCSI_PARAM_IFMARKER_EN:
+		sscanf(buf, "%d", &value);
+		if (value) {
+			printk(KERN_ERR "IFMarker wasn't negotiated to No");
+			return -EPROTO;
+		}
+		break;
+	case ISCSI_PARAM_OFMARKER_EN:
+		sscanf(buf, "%d", &value);
+		if (value) {
+			printk(KERN_ERR "OFMarker wasn't negotiated to No");
+			return -EPROTO;
+		}
+		break;
+	default:
+		return iscsi_set_param(cls_conn, param, buf, buflen);
+	}
+
+	return 0;
+}
+
+static void
+iscsi_iser_conn_get_stats(struct iscsi_cls_conn *cls_conn, struct iscsi_stats *stats)
+{
+	struct iscsi_conn *conn = cls_conn->dd_data;
+
+	stats->txdata_octets = conn->txdata_octets;
+	stats->rxdata_octets = conn->rxdata_octets;
+	stats->scsicmd_pdus = conn->scsicmd_pdus_cnt;
+	stats->dataout_pdus = conn->dataout_pdus_cnt;
+	stats->scsirsp_pdus = conn->scsirsp_pdus_cnt;
+	stats->datain_pdus = conn->datain_pdus_cnt; /* always 0 */
+	stats->r2t_pdus = conn->r2t_pdus_cnt; /* always 0 */
+	stats->tmfcmd_pdus = conn->tmfcmd_pdus_cnt;
+	stats->tmfrsp_pdus = conn->tmfrsp_pdus_cnt;
+	stats->custom_length = 4;
+	strcpy(stats->custom[0].desc, "qp_tx_queue_full");
+	stats->custom[0].value = 0; /* TB iser_conn->qp_tx_queue_full; */
+	strcpy(stats->custom[1].desc, "fmr_map_not_avail");
+	stats->custom[1].value = 0; /* TB iser_conn->fmr_map_not_avail */;
+	strcpy(stats->custom[2].desc, "eh_abort_cnt");
+	stats->custom[2].value = conn->eh_abort_cnt;
+	strcpy(stats->custom[3].desc, "fmr_unalign_cnt");
+	stats->custom[3].value = conn->fmr_unalign_cnt;
+}
+
+static struct iscsi_endpoint *
+iscsi_iser_ep_connect(struct Scsi_Host *shost, struct sockaddr *dst_addr,
+		      int non_blocking)
+{
+	int err;
+	struct iser_conn *ib_conn;
+	struct iscsi_endpoint *ep;
+
+	ep = iscsi_create_endpoint(sizeof(*ib_conn));
+	if (!ep)
+		return ERR_PTR(-ENOMEM);
+
+	ib_conn = ep->dd_data;
+	ib_conn->ep = ep;
+	iser_conn_init(ib_conn);
+
+	err = iser_connect(ib_conn, NULL, (struct sockaddr_in *)dst_addr,
+			   non_blocking);
+	if (err) {
+		iscsi_destroy_endpoint(ep);
+		return ERR_PTR(err);
+	}
+	return ep;
+}
+
+static int
+iscsi_iser_ep_poll(struct iscsi_endpoint *ep, int timeout_ms)
+{
+	struct iser_conn *ib_conn;
+	int rc;
+
+	ib_conn = ep->dd_data;
+	rc = wait_event_interruptible_timeout(ib_conn->wait,
+			     ib_conn->state == ISER_CONN_UP,
+			     msecs_to_jiffies(timeout_ms));
+
+	/* if conn establishment failed, return error code to iscsi */
+	if (!rc &&
+	    (ib_conn->state == ISER_CONN_TERMINATING ||
+	     ib_conn->state == ISER_CONN_DOWN))
+		rc = -1;
+
+	iser_err("ib conn %p rc = %d\n", ib_conn, rc);
+
+	if (rc > 0)
+		return 1; /* success, this is the equivalent of POLLOUT */
+	else if (!rc)
+		return 0; /* timeout */
+	else
+		return rc; /* signal */
+}
+
+static void
+iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep)
+{
+	struct iser_conn *ib_conn;
+
+	ib_conn = ep->dd_data;
+	if (ib_conn->iser_conn)
+		/*
+		 * Must suspend xmit path if the ep is bound to the
+		 * iscsi_conn, so we know we are not accessing the ib_conn
+		 * when we free it.
+		 *
+		 * This may not be bound if the ep poll failed.
+		 */
+		iscsi_suspend_tx(ib_conn->iser_conn->iscsi_conn);
+
+
+	iser_err("ib conn %p state %d\n",ib_conn, ib_conn->state);
+	iser_conn_terminate(ib_conn);
+}
+
+static struct scsi_host_template iscsi_iser_sht = {
+	.module                 = THIS_MODULE,
+	.name                   = "iSCSI Initiator over iSER, v." DRV_VER,
+	.queuecommand           = iscsi_queuecommand,
+	.change_queue_depth	= iscsi_change_queue_depth,
+	.sg_tablesize           = ISCSI_ISER_SG_TABLESIZE,
+	.max_sectors		= 1024,
+	.cmd_per_lun            = ISER_DEF_CMD_PER_LUN,
+	.eh_abort_handler       = iscsi_eh_abort,
+	.eh_device_reset_handler= iscsi_eh_device_reset,
+	.eh_target_reset_handler= iscsi_eh_target_reset,
+	.target_alloc		= iscsi_target_alloc,
+	.use_clustering         = DISABLE_CLUSTERING,
+	.proc_name              = "iscsi_iser",
+	.this_id                = -1,
+};
+
+static struct iscsi_transport iscsi_iser_transport = {
+	.owner                  = THIS_MODULE,
+	.name                   = "iser",
+	.caps                   = CAP_RECOVERY_L0 | CAP_MULTI_R2T,
+	.param_mask		= ISCSI_MAX_RECV_DLENGTH |
+				  ISCSI_MAX_XMIT_DLENGTH |
+				  ISCSI_HDRDGST_EN |
+				  ISCSI_DATADGST_EN |
+				  ISCSI_INITIAL_R2T_EN |
+				  ISCSI_MAX_R2T |
+				  ISCSI_IMM_DATA_EN |
+				  ISCSI_FIRST_BURST |
+				  ISCSI_MAX_BURST |
+				  ISCSI_PDU_INORDER_EN |
+				  ISCSI_DATASEQ_INORDER_EN |
+				  ISCSI_EXP_STATSN |
+				  ISCSI_PERSISTENT_PORT |
+				  ISCSI_PERSISTENT_ADDRESS |
+				  ISCSI_TARGET_NAME | ISCSI_TPGT |
+				  ISCSI_USERNAME | ISCSI_PASSWORD |
+				  ISCSI_USERNAME_IN | ISCSI_PASSWORD_IN |
+				  ISCSI_FAST_ABORT | ISCSI_ABORT_TMO |
+				  ISCSI_PING_TMO | ISCSI_RECV_TMO |
+				  ISCSI_IFACE_NAME | ISCSI_INITIATOR_NAME,
+	.host_param_mask	= ISCSI_HOST_HWADDRESS |
+				  ISCSI_HOST_NETDEV_NAME |
+				  ISCSI_HOST_INITIATOR_NAME,
+	/* session management */
+	.create_session         = iscsi_iser_session_create,
+	.destroy_session        = iscsi_iser_session_destroy,
+	/* connection management */
+	.create_conn            = iscsi_iser_conn_create,
+	.bind_conn              = iscsi_iser_conn_bind,
+	.destroy_conn           = iscsi_iser_conn_destroy,
+	.set_param              = iscsi_iser_set_param,
+	.get_conn_param		= iscsi_conn_get_param,
+	.get_session_param	= iscsi_session_get_param,
+	.start_conn             = iscsi_iser_conn_start,
+	.stop_conn              = iscsi_iser_conn_stop,
+	/* iscsi host params */
+	.get_host_param		= iscsi_host_get_param,
+	.set_host_param		= iscsi_host_set_param,
+	/* IO */
+	.send_pdu		= iscsi_conn_send_pdu,
+	.get_stats		= iscsi_iser_conn_get_stats,
+	.init_task		= iscsi_iser_task_init,
+	.xmit_task		= iscsi_iser_task_xmit,
+	.cleanup_task		= iscsi_iser_cleanup_task,
+	.alloc_pdu		= iscsi_iser_pdu_alloc,
+	/* recovery */
+	.session_recovery_timedout = iscsi_session_recovery_timedout,
+
+	.ep_connect             = iscsi_iser_ep_connect,
+	.ep_poll                = iscsi_iser_ep_poll,
+	.ep_disconnect          = iscsi_iser_ep_disconnect
+};
+
+static int __init iser_init(void)
+{
+	int err;
+
+	iser_dbg("Starting iSER datamover...\n");
+
+	if (iscsi_max_lun < 1) {
+		printk(KERN_ERR "Invalid max_lun value of %u\n", iscsi_max_lun);
+		return -EINVAL;
+	}
+
+	memset(&ig, 0, sizeof(struct iser_global));
+
+	ig.desc_cache = kmem_cache_create("iser_descriptors",
+					  sizeof (struct iser_desc),
+					  0, SLAB_HWCACHE_ALIGN,
+					  NULL);
+	if (ig.desc_cache == NULL)
+		return -ENOMEM;
+
+	/* device init is called only after the first addr resolution */
+	mutex_init(&ig.device_list_mutex);
+	INIT_LIST_HEAD(&ig.device_list);
+	mutex_init(&ig.connlist_mutex);
+	INIT_LIST_HEAD(&ig.connlist);
+
+	iscsi_iser_scsi_transport = iscsi_register_transport(
+							&iscsi_iser_transport);
+	if (!iscsi_iser_scsi_transport) {
+		iser_err("iscsi_register_transport failed\n");
+		err = -EINVAL;
+		goto register_transport_failure;
+	}
+
+	return 0;
+
+register_transport_failure:
+	kmem_cache_destroy(ig.desc_cache);
+
+	return err;
+}
+
+static void __exit iser_exit(void)
+{
+	iser_dbg("Removing iSER datamover...\n");
+	iscsi_unregister_transport(&iscsi_iser_transport);
+	kmem_cache_destroy(ig.desc_cache);
+}
+
+module_init(iser_init);
+module_exit(iser_exit);

kernel/drivers/infiniband/ulp/iser/iscsi_iser.h

@@ -0,0 +1,371 @@
+/*
+ * iSER transport for the Open iSCSI Initiator & iSER transport internals
+ *
+ * Copyright (C) 2004 Dmitry Yusupov
+ * Copyright (C) 2004 Alex Aizman
+ * Copyright (C) 2005 Mike Christie
+ * based on code maintained by open-iscsi@googlegroups.com
+ *
+ * Copyright (c) 2004, 2005, 2006 Voltaire, 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.
+ */
+#ifndef __ISCSI_ISER_H__
+#define __ISCSI_ISER_H__
+
+#include <linux/types.h>
+#include <linux/net.h>
+#include <scsi/libiscsi.h>
+#include <scsi/scsi_transport_iscsi.h>
+
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/dma-mapping.h>
+#include <linux/mutex.h>
+#include <linux/mempool.h>
+#include <linux/uio.h>
+
+#include <linux/socket.h>
+#include <linux/in.h>
+#include <linux/in6.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_fmr_pool.h>
+#include <rdma/rdma_cm.h>
+
+#define DRV_NAME	"iser"
+#define PFX		DRV_NAME ": "
+#define DRV_VER		"0.1"
+#define DRV_DATE	"May 7th, 2006"
+
+#define iser_dbg(fmt, arg...)				\
+	do {						\
+		if (iser_debug_level > 1)		\
+			printk(KERN_DEBUG PFX "%s:" fmt,\
+				__func__ , ## arg);	\
+	} while (0)
+
+#define iser_warn(fmt, arg...)				\
+	do {						\
+		if (iser_debug_level > 0)		\
+			printk(KERN_DEBUG PFX "%s:" fmt,\
+				__func__ , ## arg);	\
+	} while (0)
+
+#define iser_err(fmt, arg...)				\
+	do {						\
+		printk(KERN_ERR PFX "%s:" fmt,          \
+		       __func__ , ## arg);		\
+	} while (0)
+
+#define SHIFT_4K	12
+#define SIZE_4K	(1UL << SHIFT_4K)
+#define MASK_4K	(~(SIZE_4K-1))
+
+					/* support upto 512KB in one RDMA */
+#define ISCSI_ISER_SG_TABLESIZE         (0x80000 >> SHIFT_4K)
+#define ISER_DEF_CMD_PER_LUN		128
+
+/* QP settings */
+/* Maximal bounds on received asynchronous PDUs */
+#define ISER_MAX_RX_MISC_PDUS		4 /* NOOP_IN(2) , ASYNC_EVENT(2)   */
+
+#define ISER_MAX_TX_MISC_PDUS		6 /* NOOP_OUT(2), TEXT(1),         *
+					   * SCSI_TMFUNC(2), LOGOUT(1) */
+
+#define ISER_QP_MAX_RECV_DTOS		(ISCSI_DEF_XMIT_CMDS_MAX + \
+					ISER_MAX_RX_MISC_PDUS    +  \
+					ISER_MAX_TX_MISC_PDUS)
+
+/* the max TX (send) WR supported by the iSER QP is defined by                 *
+ * max_send_wr = T * (1 + D) + C ; D is how many inflight dataouts we expect   *
+ * to have at max for SCSI command. The tx posting & completion handling code  *
+ * supports -EAGAIN scheme where tx is suspended till the QP has room for more *
+ * send WR. D=8 comes from 64K/8K                                              */
+
+#define ISER_INFLIGHT_DATAOUTS		8
+
+#define ISER_QP_MAX_REQ_DTOS		(ISCSI_DEF_XMIT_CMDS_MAX *    \
+					(1 + ISER_INFLIGHT_DATAOUTS) + \
+					ISER_MAX_TX_MISC_PDUS        + \
+					ISER_MAX_RX_MISC_PDUS)
+
+#define ISER_VER			0x10
+#define ISER_WSV			0x08
+#define ISER_RSV			0x04
+
+struct iser_hdr {
+	u8      flags;
+	u8      rsvd[3];
+	__be32  write_stag; /* write rkey */
+	__be64  write_va;
+	__be32  read_stag;  /* read rkey */
+	__be64  read_va;
+} __attribute__((packed));
+
+
+/* Length of an object name string */
+#define ISER_OBJECT_NAME_SIZE		    64
+
+enum iser_ib_conn_state {
+	ISER_CONN_INIT,		   /* descriptor allocd, no conn          */
+	ISER_CONN_PENDING,	   /* in the process of being established */
+	ISER_CONN_UP,		   /* up and running                      */
+	ISER_CONN_TERMINATING,	   /* in the process of being terminated  */
+	ISER_CONN_DOWN,		   /* shut down                           */
+	ISER_CONN_STATES_NUM
+};
+
+enum iser_task_status {
+	ISER_TASK_STATUS_INIT = 0,
+	ISER_TASK_STATUS_STARTED,
+	ISER_TASK_STATUS_COMPLETED
+};
+
+enum iser_data_dir {
+	ISER_DIR_IN = 0,	   /* to initiator */
+	ISER_DIR_OUT,		   /* from initiator */
+	ISER_DIRS_NUM
+};
+
+struct iser_data_buf {
+	void               *buf;      /* pointer to the sg list               */
+	unsigned int       size;      /* num entries of this sg               */
+	unsigned long      data_len;  /* total data len                       */
+	unsigned int       dma_nents; /* returned by dma_map_sg               */
+	char       	   *copy_buf; /* allocated copy buf for SGs unaligned *
+	                               * for rdma which are copied            */
+	struct scatterlist sg_single; /* SG-ified clone of a non SG SC or     *
+				       * unaligned SG                         */
+  };
+
+/* fwd declarations */
+struct iser_device;
+struct iscsi_iser_conn;
+struct iscsi_iser_task;
+struct iscsi_endpoint;
+
+struct iser_mem_reg {
+	u32  lkey;
+	u32  rkey;
+	u64  va;
+	u64  len;
+	void *mem_h;
+	int  is_fmr;
+};
+
+struct iser_regd_buf {
+	struct iser_mem_reg     reg;        /* memory registration info        */
+	void                    *virt_addr;
+	struct iser_device      *device;    /* device->device for dma_unmap    */
+	u64                     dma_addr;   /* if non zero, addr for dma_unmap */
+	enum dma_data_direction direction;  /* direction for dma_unmap	       */
+	unsigned int            data_size;
+	atomic_t                ref_count;  /* refcount, freed when dec to 0   */
+};
+
+#define MAX_REGD_BUF_VECTOR_LEN	2
+
+struct iser_dto {
+	struct iscsi_iser_task *task;
+	struct iser_conn *ib_conn;
+	int                        notify_enable;
+
+	/* vector of registered buffers */
+	unsigned int               regd_vector_len;
+	struct iser_regd_buf       *regd[MAX_REGD_BUF_VECTOR_LEN];
+
+	/* offset into the registered buffer may be specified */
+	unsigned int               offset[MAX_REGD_BUF_VECTOR_LEN];
+
+	/* a smaller size may be specified, if 0, then full size is used */
+	unsigned int               used_sz[MAX_REGD_BUF_VECTOR_LEN];
+};
+
+enum iser_desc_type {
+	ISCSI_RX,
+	ISCSI_TX_CONTROL ,
+	ISCSI_TX_SCSI_COMMAND,
+	ISCSI_TX_DATAOUT
+};
+
+struct iser_desc {
+	struct iser_hdr              iser_header;
+	struct iscsi_hdr             iscsi_header;
+	struct iser_regd_buf         hdr_regd_buf;
+	void                         *data;         /* used by RX & TX_CONTROL */
+	struct iser_regd_buf         data_regd_buf; /* used by RX & TX_CONTROL */
+	enum   iser_desc_type        type;
+	struct iser_dto              dto;
+};
+
+struct iser_device {
+	struct ib_device             *ib_device;
+	struct ib_pd	             *pd;
+	struct ib_cq	             *cq;
+	struct ib_mr	             *mr;
+	struct tasklet_struct	     cq_tasklet;
+	struct list_head             ig_list; /* entry in ig devices list */
+	int                          refcount;
+};
+
+struct iser_conn {
+	struct iscsi_iser_conn       *iser_conn; /* iser conn for upcalls  */
+	struct iscsi_endpoint	     *ep;
+	enum iser_ib_conn_state	     state;	    /* rdma connection state   */
+	atomic_t		     refcount;
+	spinlock_t		     lock;	    /* used for state changes  */
+	struct iser_device           *device;       /* device context          */
+	struct rdma_cm_id            *cma_id;       /* CMA ID		       */
+	struct ib_qp	             *qp;           /* QP 		       */
+	struct ib_fmr_pool           *fmr_pool;     /* pool of IB FMRs         */
+	int                          disc_evt_flag; /* disconn event delivered */
+	wait_queue_head_t	     wait;          /* waitq for conn/disconn  */
+	atomic_t                     post_recv_buf_count; /* posted rx count   */
+	atomic_t                     post_send_buf_count; /* posted tx count   */
+	atomic_t                     unexpected_pdu_count;/* count of received *
+							   * unexpected pdus   *
+							   * not yet retired   */
+	char 			     name[ISER_OBJECT_NAME_SIZE];
+	struct iser_page_vec         *page_vec;     /* represents SG to fmr maps*
+						     * maps serialized as tx is*/
+	struct list_head	     conn_list;       /* entry in ig conn list */
+};
+
+struct iscsi_iser_conn {
+	struct iscsi_conn            *iscsi_conn;/* ptr to iscsi conn */
+	struct iser_conn             *ib_conn;   /* iSER IB conn      */
+};
+
+struct iscsi_iser_task {
+	struct iser_desc             desc;
+	struct iscsi_iser_conn	     *iser_conn;
+	enum iser_task_status 	     status;
+	int                          command_sent;  /* set if command  sent  */
+	int                          dir[ISER_DIRS_NUM];      /* set if dir use*/
+	struct iser_regd_buf         rdma_regd[ISER_DIRS_NUM];/* regd rdma buf */
+	struct iser_data_buf         data[ISER_DIRS_NUM];     /* orig. data des*/
+	struct iser_data_buf         data_copy[ISER_DIRS_NUM];/* contig. copy  */
+};
+
+struct iser_page_vec {
+	u64 *pages;
+	int length;
+	int offset;
+	int data_size;
+};
+
+struct iser_global {
+	struct mutex      device_list_mutex;/*                   */
+	struct list_head  device_list;	     /* all iSER devices */
+	struct mutex      connlist_mutex;
+	struct list_head  connlist;		/* all iSER IB connections */
+
+	struct kmem_cache *desc_cache;
+};
+
+extern struct iser_global ig;
+extern int iser_debug_level;
+
+/* allocate connection resources needed for rdma functionality */
+int iser_conn_set_full_featured_mode(struct iscsi_conn *conn);
+
+int iser_send_control(struct iscsi_conn *conn,
+		      struct iscsi_task *task);
+
+int iser_send_command(struct iscsi_conn *conn,
+		      struct iscsi_task *task);
+
+int iser_send_data_out(struct iscsi_conn *conn,
+		       struct iscsi_task *task,
+		       struct iscsi_data *hdr);
+
+void iscsi_iser_recv(struct iscsi_conn *conn,
+		     struct iscsi_hdr       *hdr,
+		     char                   *rx_data,
+		     int                    rx_data_len);
+
+void iser_conn_init(struct iser_conn *ib_conn);
+
+void iser_conn_get(struct iser_conn *ib_conn);
+
+void iser_conn_put(struct iser_conn *ib_conn);
+
+void iser_conn_terminate(struct iser_conn *ib_conn);
+
+void iser_rcv_completion(struct iser_desc *desc,
+			 unsigned long    dto_xfer_len);
+
+void iser_snd_completion(struct iser_desc *desc);
+
+void iser_task_rdma_init(struct iscsi_iser_task *task);
+
+void iser_task_rdma_finalize(struct iscsi_iser_task *task);
+
+void iser_dto_buffs_release(struct iser_dto *dto);
+
+int  iser_regd_buff_release(struct iser_regd_buf *regd_buf);
+
+void iser_reg_single(struct iser_device      *device,
+		     struct iser_regd_buf    *regd_buf,
+		     enum dma_data_direction direction);
+
+void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *task,
+				     enum iser_data_dir         cmd_dir);
+
+int  iser_reg_rdma_mem(struct iscsi_iser_task *task,
+		       enum   iser_data_dir        cmd_dir);
+
+int  iser_connect(struct iser_conn   *ib_conn,
+		  struct sockaddr_in *src_addr,
+		  struct sockaddr_in *dst_addr,
+		  int                non_blocking);
+
+int  iser_reg_page_vec(struct iser_conn     *ib_conn,
+		       struct iser_page_vec *page_vec,
+		       struct iser_mem_reg  *mem_reg);
+
+void iser_unreg_mem(struct iser_mem_reg *mem_reg);
+
+int  iser_post_recv(struct iser_desc *rx_desc);
+int  iser_post_send(struct iser_desc *tx_desc);
+
+int iser_conn_state_comp(struct iser_conn *ib_conn,
+			 enum iser_ib_conn_state comp);
+
+int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
+			    struct iser_data_buf       *data,
+			    enum   iser_data_dir       iser_dir,
+			    enum   dma_data_direction  dma_dir);
+
+void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task);
+#endif

kernel/drivers/infiniband/ulp/iser/iser_initiator.c

@@ -0,0 +1,741 @@
+/*
+ * Copyright (c) 2004, 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.
+ */
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/scatterlist.h>
+#include <linux/kfifo.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_host.h>
+
+#include "iscsi_iser.h"
+
+/* Constant PDU lengths calculations */
+#define ISER_TOTAL_HEADERS_LEN  (sizeof (struct iser_hdr) + \
+				 sizeof (struct iscsi_hdr))
+
+/* iser_dto_add_regd_buff - increments the reference count for *
+ * the registered buffer & adds it to the DTO object           */
+static void iser_dto_add_regd_buff(struct iser_dto *dto,
+				   struct iser_regd_buf *regd_buf,
+				   unsigned long use_offset,
+				   unsigned long use_size)
+{
+	int add_idx;
+
+	atomic_inc(&regd_buf->ref_count);
+
+	add_idx = dto->regd_vector_len;
+	dto->regd[add_idx] = regd_buf;
+	dto->used_sz[add_idx] = use_size;
+	dto->offset[add_idx] = use_offset;
+
+	dto->regd_vector_len++;
+}
+
+/* Register user buffer memory and initialize passive rdma
+ *  dto descriptor. Total data size is stored in
+ *  iser_task->data[ISER_DIR_IN].data_len
+ */
+static int iser_prepare_read_cmd(struct iscsi_task *task,
+				 unsigned int edtl)
+
+{
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	struct iser_regd_buf *regd_buf;
+	int err;
+	struct iser_hdr *hdr = &iser_task->desc.iser_header;
+	struct iser_data_buf *buf_in = &iser_task->data[ISER_DIR_IN];
+
+	err = iser_dma_map_task_data(iser_task,
+				     buf_in,
+				     ISER_DIR_IN,
+				     DMA_FROM_DEVICE);
+	if (err)
+		return err;
+
+	if (edtl > iser_task->data[ISER_DIR_IN].data_len) {
+		iser_err("Total data length: %ld, less than EDTL: "
+			 "%d, in READ cmd BHS itt: %d, conn: 0x%p\n",
+			 iser_task->data[ISER_DIR_IN].data_len, edtl,
+			 task->itt, iser_task->iser_conn);
+		return -EINVAL;
+	}
+
+	err = iser_reg_rdma_mem(iser_task,ISER_DIR_IN);
+	if (err) {
+		iser_err("Failed to set up Data-IN RDMA\n");
+		return err;
+	}
+	regd_buf = &iser_task->rdma_regd[ISER_DIR_IN];
+
+	hdr->flags    |= ISER_RSV;
+	hdr->read_stag = cpu_to_be32(regd_buf->reg.rkey);
+	hdr->read_va   = cpu_to_be64(regd_buf->reg.va);
+
+	iser_dbg("Cmd itt:%d READ tags RKEY:%#.4X VA:%#llX\n",
+		 task->itt, regd_buf->reg.rkey,
+		 (unsigned long long)regd_buf->reg.va);
+
+	return 0;
+}
+
+/* Register user buffer memory and initialize passive rdma
+ *  dto descriptor. Total data size is stored in
+ *  task->data[ISER_DIR_OUT].data_len
+ */
+static int
+iser_prepare_write_cmd(struct iscsi_task *task,
+		       unsigned int imm_sz,
+		       unsigned int unsol_sz,
+		       unsigned int edtl)
+{
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	struct iser_regd_buf *regd_buf;
+	int err;
+	struct iser_dto *send_dto = &iser_task->desc.dto;
+	struct iser_hdr *hdr = &iser_task->desc.iser_header;
+	struct iser_data_buf *buf_out = &iser_task->data[ISER_DIR_OUT];
+
+	err = iser_dma_map_task_data(iser_task,
+				     buf_out,
+				     ISER_DIR_OUT,
+				     DMA_TO_DEVICE);
+	if (err)
+		return err;
+
+	if (edtl > iser_task->data[ISER_DIR_OUT].data_len) {
+		iser_err("Total data length: %ld, less than EDTL: %d, "
+			 "in WRITE cmd BHS itt: %d, conn: 0x%p\n",
+			 iser_task->data[ISER_DIR_OUT].data_len,
+			 edtl, task->itt, task->conn);
+		return -EINVAL;
+	}
+
+	err = iser_reg_rdma_mem(iser_task,ISER_DIR_OUT);
+	if (err != 0) {
+		iser_err("Failed to register write cmd RDMA mem\n");
+		return err;
+	}
+
+	regd_buf = &iser_task->rdma_regd[ISER_DIR_OUT];
+
+	if (unsol_sz < edtl) {
+		hdr->flags     |= ISER_WSV;
+		hdr->write_stag = cpu_to_be32(regd_buf->reg.rkey);
+		hdr->write_va   = cpu_to_be64(regd_buf->reg.va + unsol_sz);
+
+		iser_dbg("Cmd itt:%d, WRITE tags, RKEY:%#.4X "
+			 "VA:%#llX + unsol:%d\n",
+			 task->itt, regd_buf->reg.rkey,
+			 (unsigned long long)regd_buf->reg.va, unsol_sz);
+	}
+
+	if (imm_sz > 0) {
+		iser_dbg("Cmd itt:%d, WRITE, adding imm.data sz: %d\n",
+			 task->itt, imm_sz);
+		iser_dto_add_regd_buff(send_dto,
+				       regd_buf,
+				       0,
+				       imm_sz);
+	}
+
+	return 0;
+}
+
+/**
+ * iser_post_receive_control - allocates, initializes and posts receive DTO.
+ */
+static int iser_post_receive_control(struct iscsi_conn *conn)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iser_desc     *rx_desc;
+	struct iser_regd_buf *regd_hdr;
+	struct iser_regd_buf *regd_data;
+	struct iser_dto      *recv_dto = NULL;
+	struct iser_device  *device = iser_conn->ib_conn->device;
+	int rx_data_size, err;
+	int posts, outstanding_unexp_pdus;
+
+	/* for the login sequence we must support rx of upto 8K; login is done
+	 * after conn create/bind (connect) and conn stop/bind (reconnect),
+	 * what's common for both schemes is that the connection is not started
+	 */
+	if (conn->c_stage != ISCSI_CONN_STARTED)
+		rx_data_size = ISCSI_DEF_MAX_RECV_SEG_LEN;
+	else /* FIXME till user space sets conn->max_recv_dlength correctly */
+		rx_data_size = 128;
+
+	outstanding_unexp_pdus =
+		atomic_xchg(&iser_conn->ib_conn->unexpected_pdu_count, 0);
+
+	/*
+	 * in addition to the response buffer, replace those consumed by
+	 * unexpected pdus.
+	 */
+	for (posts = 0; posts < 1 + outstanding_unexp_pdus; posts++) {
+		rx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
+		if (rx_desc == NULL) {
+			iser_err("Failed to alloc desc for post recv %d\n",
+				 posts);
+			err = -ENOMEM;
+			goto post_rx_cache_alloc_failure;
+		}
+		rx_desc->type = ISCSI_RX;
+		rx_desc->data = kmalloc(rx_data_size, GFP_NOIO);
+		if (rx_desc->data == NULL) {
+			iser_err("Failed to alloc data buf for post recv %d\n",
+				 posts);
+			err = -ENOMEM;
+			goto post_rx_kmalloc_failure;
+		}
+
+		recv_dto = &rx_desc->dto;
+		recv_dto->ib_conn = iser_conn->ib_conn;
+		recv_dto->regd_vector_len = 0;
+
+		regd_hdr = &rx_desc->hdr_regd_buf;
+		memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
+		regd_hdr->device  = device;
+		regd_hdr->virt_addr  = rx_desc; /* == &rx_desc->iser_header */
+		regd_hdr->data_size  = ISER_TOTAL_HEADERS_LEN;
+
+		iser_reg_single(device, regd_hdr, DMA_FROM_DEVICE);
+
+		iser_dto_add_regd_buff(recv_dto, regd_hdr, 0, 0);
+
+		regd_data = &rx_desc->data_regd_buf;
+		memset(regd_data, 0, sizeof(struct iser_regd_buf));
+		regd_data->device  = device;
+		regd_data->virt_addr  = rx_desc->data;
+		regd_data->data_size  = rx_data_size;
+
+		iser_reg_single(device, regd_data, DMA_FROM_DEVICE);
+
+		iser_dto_add_regd_buff(recv_dto, regd_data, 0, 0);
+
+		err = iser_post_recv(rx_desc);
+		if (err) {
+			iser_err("Failed iser_post_recv for post %d\n", posts);
+			goto post_rx_post_recv_failure;
+		}
+	}
+	/* all posts successful */
+	return 0;
+
+post_rx_post_recv_failure:
+	iser_dto_buffs_release(recv_dto);
+	kfree(rx_desc->data);
+post_rx_kmalloc_failure:
+	kmem_cache_free(ig.desc_cache, rx_desc);
+post_rx_cache_alloc_failure:
+	if (posts > 0) {
+		/*
+		 * response buffer posted, but did not replace all unexpected
+		 * pdu recv bufs. Ignore error, retry occurs next send
+		 */
+		outstanding_unexp_pdus -= (posts - 1);
+		err = 0;
+	}
+	atomic_add(outstanding_unexp_pdus,
+		   &iser_conn->ib_conn->unexpected_pdu_count);
+
+	return err;
+}
+
+/* creates a new tx descriptor and adds header regd buffer */
+static void iser_create_send_desc(struct iscsi_iser_conn *iser_conn,
+				  struct iser_desc       *tx_desc)
+{
+	struct iser_regd_buf *regd_hdr = &tx_desc->hdr_regd_buf;
+	struct iser_dto      *send_dto = &tx_desc->dto;
+
+	memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
+	regd_hdr->device  = iser_conn->ib_conn->device;
+	regd_hdr->virt_addr  = tx_desc; /* == &tx_desc->iser_header */
+	regd_hdr->data_size  = ISER_TOTAL_HEADERS_LEN;
+
+	send_dto->ib_conn         = iser_conn->ib_conn;
+	send_dto->notify_enable   = 1;
+	send_dto->regd_vector_len = 0;
+
+	memset(&tx_desc->iser_header, 0, sizeof(struct iser_hdr));
+	tx_desc->iser_header.flags = ISER_VER;
+
+	iser_dto_add_regd_buff(send_dto, regd_hdr, 0, 0);
+}
+
+/**
+ *  iser_conn_set_full_featured_mode - (iSER API)
+ */
+int iser_conn_set_full_featured_mode(struct iscsi_conn *conn)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+
+	int i;
+	/*
+	 * FIXME this value should be declared to the target during login with
+	 * the MaxOutstandingUnexpectedPDUs key when supported
+	 */
+	int initial_post_recv_bufs_num = ISER_MAX_RX_MISC_PDUS;
+
+	iser_dbg("Initially post: %d\n", initial_post_recv_bufs_num);
+
+	/* Check that there is no posted recv or send buffers left - */
+	/* they must be consumed during the login phase */
+	BUG_ON(atomic_read(&iser_conn->ib_conn->post_recv_buf_count) != 0);
+	BUG_ON(atomic_read(&iser_conn->ib_conn->post_send_buf_count) != 0);
+
+	/* Initial post receive buffers */
+	for (i = 0; i < initial_post_recv_bufs_num; i++) {
+		if (iser_post_receive_control(conn) != 0) {
+			iser_err("Failed to post recv bufs at:%d conn:0x%p\n",
+				 i, conn);
+			return -ENOMEM;
+		}
+	}
+	iser_dbg("Posted %d post recv bufs, conn:0x%p\n", i, conn);
+	return 0;
+}
+
+static int
+iser_check_xmit(struct iscsi_conn *conn, void *task)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+
+	if (atomic_read(&iser_conn->ib_conn->post_send_buf_count) ==
+	    ISER_QP_MAX_REQ_DTOS) {
+		iser_dbg("%ld can't xmit task %p\n",jiffies,task);
+		return -ENOBUFS;
+	}
+	return 0;
+}
+
+
+/**
+ * iser_send_command - send command PDU
+ */
+int iser_send_command(struct iscsi_conn *conn,
+		      struct iscsi_task *task)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	struct iser_dto *send_dto = NULL;
+	unsigned long edtl;
+	int err = 0;
+	struct iser_data_buf *data_buf;
+	struct iscsi_cmd *hdr =  (struct iscsi_cmd *)task->hdr;
+	struct scsi_cmnd *sc  =  task->sc;
+
+	if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
+		iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
+		return -EPERM;
+	}
+	if (iser_check_xmit(conn, task))
+		return -ENOBUFS;
+
+	edtl = ntohl(hdr->data_length);
+
+	/* build the tx desc regd header and add it to the tx desc dto */
+	iser_task->desc.type = ISCSI_TX_SCSI_COMMAND;
+	send_dto = &iser_task->desc.dto;
+	send_dto->task = iser_task;
+	iser_create_send_desc(iser_conn, &iser_task->desc);
+
+	if (hdr->flags & ISCSI_FLAG_CMD_READ)
+		data_buf = &iser_task->data[ISER_DIR_IN];
+	else
+		data_buf = &iser_task->data[ISER_DIR_OUT];
+
+	if (scsi_sg_count(sc)) { /* using a scatter list */
+		data_buf->buf  = scsi_sglist(sc);
+		data_buf->size = scsi_sg_count(sc);
+	}
+
+	data_buf->data_len = scsi_bufflen(sc);
+
+	if (hdr->flags & ISCSI_FLAG_CMD_READ) {
+		err = iser_prepare_read_cmd(task, edtl);
+		if (err)
+			goto send_command_error;
+	}
+	if (hdr->flags & ISCSI_FLAG_CMD_WRITE) {
+		err = iser_prepare_write_cmd(task,
+					     task->imm_count,
+				             task->imm_count +
+					     task->unsol_r2t.data_length,
+					     edtl);
+		if (err)
+			goto send_command_error;
+	}
+
+	iser_reg_single(iser_conn->ib_conn->device,
+			send_dto->regd[0], DMA_TO_DEVICE);
+
+	if (iser_post_receive_control(conn) != 0) {
+		iser_err("post_recv failed!\n");
+		err = -ENOMEM;
+		goto send_command_error;
+	}
+
+	iser_task->status = ISER_TASK_STATUS_STARTED;
+
+	err = iser_post_send(&iser_task->desc);
+	if (!err)
+		return 0;
+
+send_command_error:
+	iser_dto_buffs_release(send_dto);
+	iser_err("conn %p failed task->itt %d err %d\n",conn, task->itt, err);
+	return err;
+}
+
+/**
+ * iser_send_data_out - send data out PDU
+ */
+int iser_send_data_out(struct iscsi_conn *conn,
+		       struct iscsi_task *task,
+		       struct iscsi_data *hdr)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	struct iser_desc *tx_desc = NULL;
+	struct iser_dto *send_dto = NULL;
+	unsigned long buf_offset;
+	unsigned long data_seg_len;
+	uint32_t itt;
+	int err = 0;
+
+	if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
+		iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
+		return -EPERM;
+	}
+
+	if (iser_check_xmit(conn, task))
+		return -ENOBUFS;
+
+	itt = (__force uint32_t)hdr->itt;
+	data_seg_len = ntoh24(hdr->dlength);
+	buf_offset   = ntohl(hdr->offset);
+
+	iser_dbg("%s itt %d dseg_len %d offset %d\n",
+		 __func__,(int)itt,(int)data_seg_len,(int)buf_offset);
+
+	tx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
+	if (tx_desc == NULL) {
+		iser_err("Failed to alloc desc for post dataout\n");
+		return -ENOMEM;
+	}
+
+	tx_desc->type = ISCSI_TX_DATAOUT;
+	memcpy(&tx_desc->iscsi_header, hdr, sizeof(struct iscsi_hdr));
+
+	/* build the tx desc regd header and add it to the tx desc dto */
+	send_dto = &tx_desc->dto;
+	send_dto->task = iser_task;
+	iser_create_send_desc(iser_conn, tx_desc);
+
+	iser_reg_single(iser_conn->ib_conn->device,
+			send_dto->regd[0], DMA_TO_DEVICE);
+
+	/* all data was registered for RDMA, we can use the lkey */
+	iser_dto_add_regd_buff(send_dto,
+			       &iser_task->rdma_regd[ISER_DIR_OUT],
+			       buf_offset,
+			       data_seg_len);
+
+	if (buf_offset + data_seg_len > iser_task->data[ISER_DIR_OUT].data_len) {
+		iser_err("Offset:%ld & DSL:%ld in Data-Out "
+			 "inconsistent with total len:%ld, itt:%d\n",
+			 buf_offset, data_seg_len,
+			 iser_task->data[ISER_DIR_OUT].data_len, itt);
+		err = -EINVAL;
+		goto send_data_out_error;
+	}
+	iser_dbg("data-out itt: %d, offset: %ld, sz: %ld\n",
+		 itt, buf_offset, data_seg_len);
+
+
+	err = iser_post_send(tx_desc);
+	if (!err)
+		return 0;
+
+send_data_out_error:
+	iser_dto_buffs_release(send_dto);
+	kmem_cache_free(ig.desc_cache, tx_desc);
+	iser_err("conn %p failed err %d\n",conn, err);
+	return err;
+}
+
+int iser_send_control(struct iscsi_conn *conn,
+		      struct iscsi_task *task)
+{
+	struct iscsi_iser_conn *iser_conn = conn->dd_data;
+	struct iscsi_iser_task *iser_task = task->dd_data;
+	struct iser_desc *mdesc = &iser_task->desc;
+	struct iser_dto *send_dto = NULL;
+	unsigned long data_seg_len;
+	int err = 0;
+	struct iser_regd_buf *regd_buf;
+	struct iser_device *device;
+	unsigned char opcode;
+
+	if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
+		iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
+		return -EPERM;
+	}
+
+	if (iser_check_xmit(conn, task))
+		return -ENOBUFS;
+
+	/* build the tx desc regd header and add it to the tx desc dto */
+	mdesc->type = ISCSI_TX_CONTROL;
+	send_dto = &mdesc->dto;
+	send_dto->task = NULL;
+	iser_create_send_desc(iser_conn, mdesc);
+
+	device = iser_conn->ib_conn->device;
+
+	iser_reg_single(device, send_dto->regd[0], DMA_TO_DEVICE);
+
+	data_seg_len = ntoh24(task->hdr->dlength);
+
+	if (data_seg_len > 0) {
+		regd_buf = &mdesc->data_regd_buf;
+		memset(regd_buf, 0, sizeof(struct iser_regd_buf));
+		regd_buf->device = device;
+		regd_buf->virt_addr = task->data;
+		regd_buf->data_size = task->data_count;
+		iser_reg_single(device, regd_buf,
+				DMA_TO_DEVICE);
+		iser_dto_add_regd_buff(send_dto, regd_buf,
+				       0,
+				       data_seg_len);
+	}
+
+	opcode = task->hdr->opcode & ISCSI_OPCODE_MASK;
+
+	/* post recv buffer for response if one is expected */
+	if (!(opcode == ISCSI_OP_NOOP_OUT && task->hdr->itt == RESERVED_ITT)) {
+		if (iser_post_receive_control(conn) != 0) {
+			iser_err("post_rcv_buff failed!\n");
+			err = -ENOMEM;
+			goto send_control_error;
+		}
+	}
+
+	err = iser_post_send(mdesc);
+	if (!err)
+		return 0;
+
+send_control_error:
+	iser_dto_buffs_release(send_dto);
+	iser_err("conn %p failed err %d\n",conn, err);
+	return err;
+}
+
+/**
+ * iser_rcv_dto_completion - recv DTO completion
+ */
+void iser_rcv_completion(struct iser_desc *rx_desc,
+			 unsigned long dto_xfer_len)
+{
+	struct iser_dto *dto = &rx_desc->dto;
+	struct iscsi_iser_conn *conn = dto->ib_conn->iser_conn;
+	struct iscsi_task *task;
+	struct iscsi_iser_task *iser_task;
+	struct iscsi_hdr *hdr;
+	char   *rx_data = NULL;
+	int     rx_data_len = 0;
+	unsigned char opcode;
+
+	hdr = &rx_desc->iscsi_header;
+
+	iser_dbg("op 0x%x itt 0x%x\n", hdr->opcode,hdr->itt);
+
+	if (dto_xfer_len > ISER_TOTAL_HEADERS_LEN) { /* we have data */
+		rx_data_len = dto_xfer_len - ISER_TOTAL_HEADERS_LEN;
+		rx_data     = dto->regd[1]->virt_addr;
+		rx_data    += dto->offset[1];
+	}
+
+	opcode = hdr->opcode & ISCSI_OPCODE_MASK;
+
+	if (opcode == ISCSI_OP_SCSI_CMD_RSP) {
+		spin_lock(&conn->iscsi_conn->session->lock);
+		task = iscsi_itt_to_ctask(conn->iscsi_conn, hdr->itt);
+		if (task)
+			__iscsi_get_task(task);
+		spin_unlock(&conn->iscsi_conn->session->lock);
+
+		if (!task)
+			iser_err("itt can't be matched to task!!! "
+				 "conn %p opcode %d itt %d\n",
+				 conn->iscsi_conn, opcode, hdr->itt);
+		else {
+			iser_task = task->dd_data;
+			iser_dbg("itt %d task %p\n",hdr->itt, task);
+			iser_task->status = ISER_TASK_STATUS_COMPLETED;
+			iser_task_rdma_finalize(iser_task);
+			iscsi_put_task(task);
+		}
+	}
+	iser_dto_buffs_release(dto);
+
+	iscsi_iser_recv(conn->iscsi_conn, hdr, rx_data, rx_data_len);
+
+	kfree(rx_desc->data);
+	kmem_cache_free(ig.desc_cache, rx_desc);
+
+	/* decrementing conn->post_recv_buf_count only --after-- freeing the   *
+	 * task eliminates the need to worry on tasks which are completed in   *
+	 * parallel to the execution of iser_conn_term. So the code that waits *
+	 * for the posted rx bufs refcount to become zero handles everything   */
+	atomic_dec(&conn->ib_conn->post_recv_buf_count);
+
+	/*
+	 * if an unexpected PDU was received then the recv wr consumed must
+	 * be replaced, this is done in the next send of a control-type PDU
+	 */
+	if (opcode == ISCSI_OP_NOOP_IN && hdr->itt == RESERVED_ITT) {
+		/* nop-in with itt = 0xffffffff */
+		atomic_inc(&conn->ib_conn->unexpected_pdu_count);
+	}
+	else if (opcode == ISCSI_OP_ASYNC_EVENT) {
+		/* asyncronous message */
+		atomic_inc(&conn->ib_conn->unexpected_pdu_count);
+	}
+	/* a reject PDU consumes the recv buf posted for the response */
+}
+
+void iser_snd_completion(struct iser_desc *tx_desc)
+{
+	struct iser_dto        *dto = &tx_desc->dto;
+	struct iser_conn       *ib_conn = dto->ib_conn;
+	struct iscsi_iser_conn *iser_conn = ib_conn->iser_conn;
+	struct iscsi_conn      *conn = iser_conn->iscsi_conn;
+	struct iscsi_task *task;
+	int resume_tx = 0;
+
+	iser_dbg("Initiator, Data sent dto=0x%p\n", dto);
+
+	iser_dto_buffs_release(dto);
+
+	if (tx_desc->type == ISCSI_TX_DATAOUT)
+		kmem_cache_free(ig.desc_cache, tx_desc);
+
+	if (atomic_read(&iser_conn->ib_conn->post_send_buf_count) ==
+	    ISER_QP_MAX_REQ_DTOS)
+		resume_tx = 1;
+
+	atomic_dec(&ib_conn->post_send_buf_count);
+
+	if (resume_tx) {
+		iser_dbg("%ld resuming tx\n",jiffies);
+		iscsi_conn_queue_work(conn);
+	}
+
+	if (tx_desc->type == ISCSI_TX_CONTROL) {
+		/* this arithmetic is legal by libiscsi dd_data allocation */
+		task = (void *) ((long)(void *)tx_desc -
+				  sizeof(struct iscsi_task));
+		if (task->hdr->itt == RESERVED_ITT)
+			iscsi_put_task(task);
+	}
+}
+
+void iser_task_rdma_init(struct iscsi_iser_task *iser_task)
+
+{
+	iser_task->status = ISER_TASK_STATUS_INIT;
+
+	iser_task->dir[ISER_DIR_IN] = 0;
+	iser_task->dir[ISER_DIR_OUT] = 0;
+
+	iser_task->data[ISER_DIR_IN].data_len  = 0;
+	iser_task->data[ISER_DIR_OUT].data_len = 0;
+
+	memset(&iser_task->rdma_regd[ISER_DIR_IN], 0,
+	       sizeof(struct iser_regd_buf));
+	memset(&iser_task->rdma_regd[ISER_DIR_OUT], 0,
+	       sizeof(struct iser_regd_buf));
+}
+
+void iser_task_rdma_finalize(struct iscsi_iser_task *iser_task)
+{
+	int deferred;
+	int is_rdma_aligned = 1;
+	struct iser_regd_buf *regd;
+
+	/* if we were reading, copy back to unaligned sglist,
+	 * anyway dma_unmap and free the copy
+	 */
+	if (iser_task->data_copy[ISER_DIR_IN].copy_buf != NULL) {
+		is_rdma_aligned = 0;
+		iser_finalize_rdma_unaligned_sg(iser_task, ISER_DIR_IN);
+	}
+	if (iser_task->data_copy[ISER_DIR_OUT].copy_buf != NULL) {
+		is_rdma_aligned = 0;
+		iser_finalize_rdma_unaligned_sg(iser_task, ISER_DIR_OUT);
+	}
+
+	if (iser_task->dir[ISER_DIR_IN]) {
+		regd = &iser_task->rdma_regd[ISER_DIR_IN];
+		deferred = iser_regd_buff_release(regd);
+		if (deferred) {
+			iser_err("%d references remain for BUF-IN rdma reg\n",
+				 atomic_read(&regd->ref_count));
+		}
+	}
+
+	if (iser_task->dir[ISER_DIR_OUT]) {
+		regd = &iser_task->rdma_regd[ISER_DIR_OUT];
+		deferred = iser_regd_buff_release(regd);
+		if (deferred) {
+			iser_err("%d references remain for BUF-OUT rdma reg\n",
+				 atomic_read(&regd->ref_count));
+		}
+	}
+
+       /* if the data was unaligned, it was already unmapped and then copied */
+       if (is_rdma_aligned)
+		iser_dma_unmap_task_data(iser_task);
+}
+
+void iser_dto_buffs_release(struct iser_dto *dto)
+{
+	int i;
+
+	for (i = 0; i < dto->regd_vector_len; i++)
+		iser_regd_buff_release(dto->regd[i]);
+}
+

kernel/drivers/infiniband/ulp/iser/iser_memory.c

@@ -0,0 +1,482 @@
+/*
+ * Copyright (c) 2004, 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.
+ */
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/highmem.h>
+#include <linux/scatterlist.h>
+
+#include "iscsi_iser.h"
+
+#define ISER_KMALLOC_THRESHOLD 0x20000 /* 128K - kmalloc limit */
+
+/**
+ * Decrements the reference count for the
+ * registered buffer & releases it
+ *
+ * returns 0 if released, 1 if deferred
+ */
+int iser_regd_buff_release(struct iser_regd_buf *regd_buf)
+{
+	struct ib_device *dev;
+
+	if ((atomic_read(&regd_buf->ref_count) == 0) ||
+	    atomic_dec_and_test(&regd_buf->ref_count)) {
+		/* if we used the dma mr, unreg is just NOP */
+		if (regd_buf->reg.is_fmr)
+			iser_unreg_mem(&regd_buf->reg);
+
+		if (regd_buf->dma_addr) {
+			dev = regd_buf->device->ib_device;
+			ib_dma_unmap_single(dev,
+					 regd_buf->dma_addr,
+					 regd_buf->data_size,
+					 regd_buf->direction);
+		}
+		/* else this regd buf is associated with task which we */
+		/* dma_unmap_single/sg later */
+		return 0;
+	} else {
+		iser_dbg("Release deferred, regd.buff: 0x%p\n", regd_buf);
+		return 1;
+	}
+}
+
+/**
+ * iser_reg_single - fills registered buffer descriptor with
+ *		     registration information
+ */
+void iser_reg_single(struct iser_device *device,
+		     struct iser_regd_buf *regd_buf,
+		     enum dma_data_direction direction)
+{
+	u64 dma_addr;
+
+	dma_addr = ib_dma_map_single(device->ib_device,
+				     regd_buf->virt_addr,
+				     regd_buf->data_size, direction);
+	BUG_ON(ib_dma_mapping_error(device->ib_device, dma_addr));
+
+	regd_buf->reg.lkey = device->mr->lkey;
+	regd_buf->reg.len  = regd_buf->data_size;
+	regd_buf->reg.va   = dma_addr;
+	regd_buf->reg.is_fmr = 0;
+
+	regd_buf->dma_addr  = dma_addr;
+	regd_buf->direction = direction;
+}
+
+/**
+ * iser_start_rdma_unaligned_sg
+ */
+static int iser_start_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
+					enum iser_data_dir cmd_dir)
+{
+	int dma_nents;
+	struct ib_device *dev;
+	char *mem = NULL;
+	struct iser_data_buf *data = &iser_task->data[cmd_dir];
+	unsigned long  cmd_data_len = data->data_len;
+
+	if (cmd_data_len > ISER_KMALLOC_THRESHOLD)
+		mem = (void *)__get_free_pages(GFP_NOIO,
+		      ilog2(roundup_pow_of_two(cmd_data_len)) - PAGE_SHIFT);
+	else
+		mem = kmalloc(cmd_data_len, GFP_NOIO);
+
+	if (mem == NULL) {
+		iser_err("Failed to allocate mem size %d %d for copying sglist\n",
+			 data->size,(int)cmd_data_len);
+		return -ENOMEM;
+	}
+
+	if (cmd_dir == ISER_DIR_OUT) {
+		/* copy the unaligned sg the buffer which is used for RDMA */
+		struct scatterlist *sgl = (struct scatterlist *)data->buf;
+		struct scatterlist *sg;
+		int i;
+		char *p, *from;
+
+		p = mem;
+		for_each_sg(sgl, sg, data->size, i) {
+			from = kmap_atomic(sg_page(sg), KM_USER0);
+			memcpy(p,
+			       from + sg->offset,
+			       sg->length);
+			kunmap_atomic(from, KM_USER0);
+			p += sg->length;
+		}
+	}
+
+	sg_init_one(&iser_task->data_copy[cmd_dir].sg_single, mem, cmd_data_len);
+	iser_task->data_copy[cmd_dir].buf  =
+		&iser_task->data_copy[cmd_dir].sg_single;
+	iser_task->data_copy[cmd_dir].size = 1;
+
+	iser_task->data_copy[cmd_dir].copy_buf  = mem;
+
+	dev = iser_task->iser_conn->ib_conn->device->ib_device;
+	dma_nents = ib_dma_map_sg(dev,
+				  &iser_task->data_copy[cmd_dir].sg_single,
+				  1,
+				  (cmd_dir == ISER_DIR_OUT) ?
+				  DMA_TO_DEVICE : DMA_FROM_DEVICE);
+	BUG_ON(dma_nents == 0);
+
+	iser_task->data_copy[cmd_dir].dma_nents = dma_nents;
+	return 0;
+}
+
+/**
+ * iser_finalize_rdma_unaligned_sg
+ */
+void iser_finalize_rdma_unaligned_sg(struct iscsi_iser_task *iser_task,
+				     enum iser_data_dir         cmd_dir)
+{
+	struct ib_device *dev;
+	struct iser_data_buf *mem_copy;
+	unsigned long  cmd_data_len;
+
+	dev = iser_task->iser_conn->ib_conn->device->ib_device;
+	mem_copy = &iser_task->data_copy[cmd_dir];
+
+	ib_dma_unmap_sg(dev, &mem_copy->sg_single, 1,
+			(cmd_dir == ISER_DIR_OUT) ?
+			DMA_TO_DEVICE : DMA_FROM_DEVICE);
+
+	if (cmd_dir == ISER_DIR_IN) {
+		char *mem;
+		struct scatterlist *sgl, *sg;
+		unsigned char *p, *to;
+		unsigned int sg_size;
+		int i;
+
+		/* copy back read RDMA to unaligned sg */
+		mem	= mem_copy->copy_buf;
+
+		sgl	= (struct scatterlist *)iser_task->data[ISER_DIR_IN].buf;
+		sg_size = iser_task->data[ISER_DIR_IN].size;
+
+		p = mem;
+		for_each_sg(sgl, sg, sg_size, i) {
+			to = kmap_atomic(sg_page(sg), KM_SOFTIRQ0);
+			memcpy(to + sg->offset,
+			       p,
+			       sg->length);
+			kunmap_atomic(to, KM_SOFTIRQ0);
+			p += sg->length;
+		}
+	}
+
+	cmd_data_len = iser_task->data[cmd_dir].data_len;
+
+	if (cmd_data_len > ISER_KMALLOC_THRESHOLD)
+		free_pages((unsigned long)mem_copy->copy_buf,
+			   ilog2(roundup_pow_of_two(cmd_data_len)) - PAGE_SHIFT);
+	else
+		kfree(mem_copy->copy_buf);
+
+	mem_copy->copy_buf = NULL;
+}
+
+#define IS_4K_ALIGNED(addr)	((((unsigned long)addr) & ~MASK_4K) == 0)
+
+/**
+ * iser_sg_to_page_vec - Translates scatterlist entries to physical addresses
+ * and returns the length of resulting physical address array (may be less than
+ * the original due to possible compaction).
+ *
+ * we build a "page vec" under the assumption that the SG meets the RDMA
+ * alignment requirements. Other then the first and last SG elements, all
+ * the "internal" elements can be compacted into a list whose elements are
+ * dma addresses of physical pages. The code supports also the weird case
+ * where --few fragments of the same page-- are present in the SG as
+ * consecutive elements. Also, it handles one entry SG.
+ */
+
+static int iser_sg_to_page_vec(struct iser_data_buf *data,
+			       struct iser_page_vec *page_vec,
+			       struct ib_device *ibdev)
+{
+	struct scatterlist *sg, *sgl = (struct scatterlist *)data->buf;
+	u64 start_addr, end_addr, page, chunk_start = 0;
+	unsigned long total_sz = 0;
+	unsigned int dma_len;
+	int i, new_chunk, cur_page, last_ent = data->dma_nents - 1;
+
+	/* compute the offset of first element */
+	page_vec->offset = (u64) sgl[0].offset & ~MASK_4K;
+
+	new_chunk = 1;
+	cur_page  = 0;
+	for_each_sg(sgl, sg, data->dma_nents, i) {
+		start_addr = ib_sg_dma_address(ibdev, sg);
+		if (new_chunk)
+			chunk_start = start_addr;
+		dma_len = ib_sg_dma_len(ibdev, sg);
+		end_addr = start_addr + dma_len;
+		total_sz += dma_len;
+
+		/* collect page fragments until aligned or end of SG list */
+		if (!IS_4K_ALIGNED(end_addr) && i < last_ent) {
+			new_chunk = 0;
+			continue;
+		}
+		new_chunk = 1;
+
+		/* address of the first page in the contiguous chunk;
+		   masking relevant for the very first SG entry,
+		   which might be unaligned */
+		page = chunk_start & MASK_4K;
+		do {
+			page_vec->pages[cur_page++] = page;
+			page += SIZE_4K;
+		} while (page < end_addr);
+	}
+
+	page_vec->data_size = total_sz;
+	iser_dbg("page_vec->data_size:%d cur_page %d\n", page_vec->data_size,cur_page);
+	return cur_page;
+}
+
+
+/**
+ * iser_data_buf_aligned_len - Tries to determine the maximal correctly aligned
+ * for RDMA sub-list of a scatter-gather list of memory buffers, and  returns
+ * the number of entries which are aligned correctly. Supports the case where
+ * consecutive SG elements are actually fragments of the same physcial page.
+ */
+static int iser_data_buf_aligned_len(struct iser_data_buf *data,
+				      struct ib_device *ibdev)
+{
+	struct scatterlist *sgl, *sg, *next_sg = NULL;
+	u64 start_addr, end_addr;
+	int i, ret_len, start_check = 0;
+
+	if (data->dma_nents == 1)
+		return 1;
+
+	sgl = (struct scatterlist *)data->buf;
+	start_addr  = ib_sg_dma_address(ibdev, sgl);
+
+	for_each_sg(sgl, sg, data->dma_nents, i) {
+		if (start_check && !IS_4K_ALIGNED(start_addr))
+			break;
+
+		next_sg = sg_next(sg);
+		if (!next_sg)
+			break;
+
+		end_addr    = start_addr + ib_sg_dma_len(ibdev, sg);
+		start_addr  = ib_sg_dma_address(ibdev, next_sg);
+
+		if (end_addr == start_addr) {
+			start_check = 0;
+			continue;
+		} else
+			start_check = 1;
+
+		if (!IS_4K_ALIGNED(end_addr))
+			break;
+	}
+	ret_len = (next_sg) ? i : i+1;
+	iser_dbg("Found %d aligned entries out of %d in sg:0x%p\n",
+		 ret_len, data->dma_nents, data);
+	return ret_len;
+}
+
+static void iser_data_buf_dump(struct iser_data_buf *data,
+			       struct ib_device *ibdev)
+{
+	struct scatterlist *sgl = (struct scatterlist *)data->buf;
+	struct scatterlist *sg;
+	int i;
+
+	if (iser_debug_level == 0)
+		return;
+
+	for_each_sg(sgl, sg, data->dma_nents, i)
+		iser_warn("sg[%d] dma_addr:0x%lX page:0x%p "
+			 "off:0x%x sz:0x%x dma_len:0x%x\n",
+			 i, (unsigned long)ib_sg_dma_address(ibdev, sg),
+			 sg_page(sg), sg->offset,
+			 sg->length, ib_sg_dma_len(ibdev, sg));
+}
+
+static void iser_dump_page_vec(struct iser_page_vec *page_vec)
+{
+	int i;
+
+	iser_err("page vec length %d data size %d\n",
+		 page_vec->length, page_vec->data_size);
+	for (i = 0; i < page_vec->length; i++)
+		iser_err("%d %lx\n",i,(unsigned long)page_vec->pages[i]);
+}
+
+static void iser_page_vec_build(struct iser_data_buf *data,
+				struct iser_page_vec *page_vec,
+				struct ib_device *ibdev)
+{
+	int page_vec_len = 0;
+
+	page_vec->length = 0;
+	page_vec->offset = 0;
+
+	iser_dbg("Translating sg sz: %d\n", data->dma_nents);
+	page_vec_len = iser_sg_to_page_vec(data, page_vec, ibdev);
+	iser_dbg("sg len %d page_vec_len %d\n", data->dma_nents,page_vec_len);
+
+	page_vec->length = page_vec_len;
+
+	if (page_vec_len * SIZE_4K < page_vec->data_size) {
+		iser_err("page_vec too short to hold this SG\n");
+		iser_data_buf_dump(data, ibdev);
+		iser_dump_page_vec(page_vec);
+		BUG();
+	}
+}
+
+int iser_dma_map_task_data(struct iscsi_iser_task *iser_task,
+			    struct iser_data_buf *data,
+			    enum iser_data_dir iser_dir,
+			    enum dma_data_direction dma_dir)
+{
+	struct ib_device *dev;
+
+	iser_task->dir[iser_dir] = 1;
+	dev = iser_task->iser_conn->ib_conn->device->ib_device;
+
+	data->dma_nents = ib_dma_map_sg(dev, data->buf, data->size, dma_dir);
+	if (data->dma_nents == 0) {
+		iser_err("dma_map_sg failed!!!\n");
+		return -EINVAL;
+	}
+	return 0;
+}
+
+void iser_dma_unmap_task_data(struct iscsi_iser_task *iser_task)
+{
+	struct ib_device *dev;
+	struct iser_data_buf *data;
+
+	dev = iser_task->iser_conn->ib_conn->device->ib_device;
+
+	if (iser_task->dir[ISER_DIR_IN]) {
+		data = &iser_task->data[ISER_DIR_IN];
+		ib_dma_unmap_sg(dev, data->buf, data->size, DMA_FROM_DEVICE);
+	}
+
+	if (iser_task->dir[ISER_DIR_OUT]) {
+		data = &iser_task->data[ISER_DIR_OUT];
+		ib_dma_unmap_sg(dev, data->buf, data->size, DMA_TO_DEVICE);
+	}
+}
+
+/**
+ * iser_reg_rdma_mem - Registers memory intended for RDMA,
+ * obtaining rkey and va
+ *
+ * returns 0 on success, errno code on failure
+ */
+int iser_reg_rdma_mem(struct iscsi_iser_task *iser_task,
+		      enum   iser_data_dir        cmd_dir)
+{
+	struct iscsi_conn    *iscsi_conn = iser_task->iser_conn->iscsi_conn;
+	struct iser_conn     *ib_conn = iser_task->iser_conn->ib_conn;
+	struct iser_device   *device = ib_conn->device;
+	struct ib_device     *ibdev = device->ib_device;
+	struct iser_data_buf *mem = &iser_task->data[cmd_dir];
+	struct iser_regd_buf *regd_buf;
+	int aligned_len;
+	int err;
+	int i;
+	struct scatterlist *sg;
+
+	regd_buf = &iser_task->rdma_regd[cmd_dir];
+
+	aligned_len = iser_data_buf_aligned_len(mem, ibdev);
+	if (aligned_len != mem->dma_nents) {
+		iscsi_conn->fmr_unalign_cnt++;
+		iser_warn("rdma alignment violation %d/%d aligned\n",
+			 aligned_len, mem->size);
+		iser_data_buf_dump(mem, ibdev);
+
+		/* unmap the command data before accessing it */
+		iser_dma_unmap_task_data(iser_task);
+
+		/* allocate copy buf, if we are writing, copy the */
+		/* unaligned scatterlist, dma map the copy        */
+		if (iser_start_rdma_unaligned_sg(iser_task, cmd_dir) != 0)
+				return -ENOMEM;
+		mem = &iser_task->data_copy[cmd_dir];
+	}
+
+	/* if there a single dma entry, FMR is not needed */
+	if (mem->dma_nents == 1) {
+		sg = (struct scatterlist *)mem->buf;
+
+		regd_buf->reg.lkey = device->mr->lkey;
+		regd_buf->reg.rkey = device->mr->rkey;
+		regd_buf->reg.len  = ib_sg_dma_len(ibdev, &sg[0]);
+		regd_buf->reg.va   = ib_sg_dma_address(ibdev, &sg[0]);
+		regd_buf->reg.is_fmr = 0;
+
+		iser_dbg("PHYSICAL Mem.register: lkey: 0x%08X rkey: 0x%08X  "
+			 "va: 0x%08lX sz: %ld]\n",
+			 (unsigned int)regd_buf->reg.lkey,
+			 (unsigned int)regd_buf->reg.rkey,
+			 (unsigned long)regd_buf->reg.va,
+			 (unsigned long)regd_buf->reg.len);
+	} else { /* use FMR for multiple dma entries */
+		iser_page_vec_build(mem, ib_conn->page_vec, ibdev);
+		err = iser_reg_page_vec(ib_conn, ib_conn->page_vec, &regd_buf->reg);
+		if (err) {
+			iser_data_buf_dump(mem, ibdev);
+			iser_err("mem->dma_nents = %d (dlength = 0x%x)\n",
+				 mem->dma_nents,
+				 ntoh24(iser_task->desc.iscsi_header.dlength));
+			iser_err("page_vec: data_size = 0x%x, length = %d, offset = 0x%x\n",
+				 ib_conn->page_vec->data_size, ib_conn->page_vec->length,
+				 ib_conn->page_vec->offset);
+			for (i=0 ; i<ib_conn->page_vec->length ; i++)
+				iser_err("page_vec[%d] = 0x%llx\n", i,
+					 (unsigned long long) ib_conn->page_vec->pages[i]);
+			return err;
+		}
+	}
+
+	/* take a reference on this regd buf such that it will not be released *
+	 * (eg in send dto completion) before we get the scsi response         */
+	atomic_inc(&regd_buf->ref_count);
+	return 0;
+}

kernel/drivers/infiniband/ulp/iser/iser_verbs.c

@@ -0,0 +1,810 @@
+/*
+ * Copyright (c) 2004, 2005, 2006 Voltaire, 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.
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+
+#include "iscsi_iser.h"
+
+#define ISCSI_ISER_MAX_CONN	8
+#define ISER_MAX_CQ_LEN		((ISER_QP_MAX_RECV_DTOS + \
+				ISER_QP_MAX_REQ_DTOS) *   \
+				 ISCSI_ISER_MAX_CONN)
+
+static void iser_cq_tasklet_fn(unsigned long data);
+static void iser_cq_callback(struct ib_cq *cq, void *cq_context);
+
+static void iser_cq_event_callback(struct ib_event *cause, void *context)
+{
+	iser_err("got cq event %d \n", cause->event);
+}
+
+static void iser_qp_event_callback(struct ib_event *cause, void *context)
+{
+	iser_err("got qp event %d\n",cause->event);
+}
+
+/**
+ * iser_create_device_ib_res - creates Protection Domain (PD), Completion
+ * Queue (CQ), DMA Memory Region (DMA MR) with the device associated with
+ * the adapator.
+ *
+ * returns 0 on success, -1 on failure
+ */
+static int iser_create_device_ib_res(struct iser_device *device)
+{
+	device->pd = ib_alloc_pd(device->ib_device);
+	if (IS_ERR(device->pd))
+		goto pd_err;
+
+	device->cq = ib_create_cq(device->ib_device,
+				  iser_cq_callback,
+				  iser_cq_event_callback,
+				  (void *)device,
+				  ISER_MAX_CQ_LEN, 0);
+	if (IS_ERR(device->cq))
+		goto cq_err;
+
+	if (ib_req_notify_cq(device->cq, IB_CQ_NEXT_COMP))
+		goto cq_arm_err;
+
+	tasklet_init(&device->cq_tasklet,
+		     iser_cq_tasklet_fn,
+		     (unsigned long)device);
+
+	device->mr = ib_get_dma_mr(device->pd, IB_ACCESS_LOCAL_WRITE |
+				   IB_ACCESS_REMOTE_WRITE |
+				   IB_ACCESS_REMOTE_READ);
+	if (IS_ERR(device->mr))
+		goto dma_mr_err;
+
+	return 0;
+
+dma_mr_err:
+	tasklet_kill(&device->cq_tasklet);
+cq_arm_err:
+	ib_destroy_cq(device->cq);
+cq_err:
+	ib_dealloc_pd(device->pd);
+pd_err:
+	iser_err("failed to allocate an IB resource\n");
+	return -1;
+}
+
+/**
+ * iser_free_device_ib_res - destroy/dealloc/dereg the DMA MR,
+ * CQ and PD created with the device associated with the adapator.
+ */
+static void iser_free_device_ib_res(struct iser_device *device)
+{
+	BUG_ON(device->mr == NULL);
+
+	tasklet_kill(&device->cq_tasklet);
+
+	(void)ib_dereg_mr(device->mr);
+	(void)ib_destroy_cq(device->cq);
+	(void)ib_dealloc_pd(device->pd);
+
+	device->mr = NULL;
+	device->cq = NULL;
+	device->pd = NULL;
+}
+
+/**
+ * iser_create_ib_conn_res - Creates FMR pool and Queue-Pair (QP)
+ *
+ * returns 0 on success, -1 on failure
+ */
+static int iser_create_ib_conn_res(struct iser_conn *ib_conn)
+{
+	struct iser_device	*device;
+	struct ib_qp_init_attr	init_attr;
+	int			ret;
+	struct ib_fmr_pool_param params;
+
+	BUG_ON(ib_conn->device == NULL);
+
+	device = ib_conn->device;
+
+	ib_conn->page_vec = kmalloc(sizeof(struct iser_page_vec) +
+				    (sizeof(u64) * (ISCSI_ISER_SG_TABLESIZE +1)),
+				    GFP_KERNEL);
+	if (!ib_conn->page_vec) {
+		ret = -ENOMEM;
+		goto alloc_err;
+	}
+	ib_conn->page_vec->pages = (u64 *) (ib_conn->page_vec + 1);
+
+	params.page_shift        = SHIFT_4K;
+	/* when the first/last SG element are not start/end *
+	 * page aligned, the map whould be of N+1 pages     */
+	params.max_pages_per_fmr = ISCSI_ISER_SG_TABLESIZE + 1;
+	/* make the pool size twice the max number of SCSI commands *
+	 * the ML is expected to queue, watermark for unmap at 50%  */
+	params.pool_size	 = ISCSI_DEF_XMIT_CMDS_MAX * 2;
+	params.dirty_watermark	 = ISCSI_DEF_XMIT_CMDS_MAX;
+	params.cache		 = 0;
+	params.flush_function	 = NULL;
+	params.access		 = (IB_ACCESS_LOCAL_WRITE  |
+				    IB_ACCESS_REMOTE_WRITE |
+				    IB_ACCESS_REMOTE_READ);
+
+	ib_conn->fmr_pool = ib_create_fmr_pool(device->pd, &params);
+	if (IS_ERR(ib_conn->fmr_pool)) {
+		ret = PTR_ERR(ib_conn->fmr_pool);
+		goto fmr_pool_err;
+	}
+
+	memset(&init_attr, 0, sizeof init_attr);
+
+	init_attr.event_handler = iser_qp_event_callback;
+	init_attr.qp_context	= (void *)ib_conn;
+	init_attr.send_cq	= device->cq;
+	init_attr.recv_cq	= device->cq;
+	init_attr.cap.max_send_wr  = ISER_QP_MAX_REQ_DTOS;
+	init_attr.cap.max_recv_wr  = ISER_QP_MAX_RECV_DTOS;
+	init_attr.cap.max_send_sge = MAX_REGD_BUF_VECTOR_LEN;
+	init_attr.cap.max_recv_sge = 2;
+	init_attr.sq_sig_type	= IB_SIGNAL_REQ_WR;
+	init_attr.qp_type	= IB_QPT_RC;
+
+	ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr);
+	if (ret)
+		goto qp_err;
+
+	ib_conn->qp = ib_conn->cma_id->qp;
+	iser_err("setting conn %p cma_id %p: fmr_pool %p qp %p\n",
+		 ib_conn, ib_conn->cma_id,
+		 ib_conn->fmr_pool, ib_conn->cma_id->qp);
+	return ret;
+
+qp_err:
+	(void)ib_destroy_fmr_pool(ib_conn->fmr_pool);
+fmr_pool_err:
+	kfree(ib_conn->page_vec);
+alloc_err:
+	iser_err("unable to alloc mem or create resource, err %d\n", ret);
+	return ret;
+}
+
+/**
+ * releases the FMR pool, QP and CMA ID objects, returns 0 on success,
+ * -1 on failure
+ */
+static int iser_free_ib_conn_res(struct iser_conn *ib_conn)
+{
+	BUG_ON(ib_conn == NULL);
+
+	iser_err("freeing conn %p cma_id %p fmr pool %p qp %p\n",
+		 ib_conn, ib_conn->cma_id,
+		 ib_conn->fmr_pool, ib_conn->qp);
+
+	/* qp is created only once both addr & route are resolved */
+	if (ib_conn->fmr_pool != NULL)
+		ib_destroy_fmr_pool(ib_conn->fmr_pool);
+
+	if (ib_conn->qp != NULL)
+		rdma_destroy_qp(ib_conn->cma_id);
+
+	if (ib_conn->cma_id != NULL)
+		rdma_destroy_id(ib_conn->cma_id);
+
+	ib_conn->fmr_pool = NULL;
+	ib_conn->qp	  = NULL;
+	ib_conn->cma_id   = NULL;
+	kfree(ib_conn->page_vec);
+
+	return 0;
+}
+
+/**
+ * based on the resolved device node GUID see if there already allocated
+ * device for this device. If there's no such, create one.
+ */
+static
+struct iser_device *iser_device_find_by_ib_device(struct rdma_cm_id *cma_id)
+{
+	struct iser_device *device;
+
+	mutex_lock(&ig.device_list_mutex);
+
+	list_for_each_entry(device, &ig.device_list, ig_list)
+		/* find if there's a match using the node GUID */
+		if (device->ib_device->node_guid == cma_id->device->node_guid)
+			goto inc_refcnt;
+
+	device = kzalloc(sizeof *device, GFP_KERNEL);
+	if (device == NULL)
+		goto out;
+
+	/* assign this device to the device */
+	device->ib_device = cma_id->device;
+	/* init the device and link it into ig device list */
+	if (iser_create_device_ib_res(device)) {
+		kfree(device);
+		device = NULL;
+		goto out;
+	}
+	list_add(&device->ig_list, &ig.device_list);
+
+inc_refcnt:
+	device->refcount++;
+out:
+	mutex_unlock(&ig.device_list_mutex);
+	return device;
+}
+
+/* if there's no demand for this device, release it */
+static void iser_device_try_release(struct iser_device *device)
+{
+	mutex_lock(&ig.device_list_mutex);
+	device->refcount--;
+	iser_err("device %p refcount %d\n",device,device->refcount);
+	if (!device->refcount) {
+		iser_free_device_ib_res(device);
+		list_del(&device->ig_list);
+		kfree(device);
+	}
+	mutex_unlock(&ig.device_list_mutex);
+}
+
+int iser_conn_state_comp(struct iser_conn *ib_conn,
+			enum iser_ib_conn_state comp)
+{
+	int ret;
+
+	spin_lock_bh(&ib_conn->lock);
+	ret = (ib_conn->state == comp);
+	spin_unlock_bh(&ib_conn->lock);
+	return ret;
+}
+
+static int iser_conn_state_comp_exch(struct iser_conn *ib_conn,
+				     enum iser_ib_conn_state comp,
+				     enum iser_ib_conn_state exch)
+{
+	int ret;
+
+	spin_lock_bh(&ib_conn->lock);
+	if ((ret = (ib_conn->state == comp)))
+		ib_conn->state = exch;
+	spin_unlock_bh(&ib_conn->lock);
+	return ret;
+}
+
+/**
+ * Frees all conn objects and deallocs conn descriptor
+ */
+static void iser_conn_release(struct iser_conn *ib_conn)
+{
+	struct iser_device  *device = ib_conn->device;
+
+	BUG_ON(ib_conn->state != ISER_CONN_DOWN);
+
+	mutex_lock(&ig.connlist_mutex);
+	list_del(&ib_conn->conn_list);
+	mutex_unlock(&ig.connlist_mutex);
+
+	iser_free_ib_conn_res(ib_conn);
+	ib_conn->device = NULL;
+	/* on EVENT_ADDR_ERROR there's no device yet for this conn */
+	if (device != NULL)
+		iser_device_try_release(device);
+	if (ib_conn->iser_conn)
+		ib_conn->iser_conn->ib_conn = NULL;
+	iscsi_destroy_endpoint(ib_conn->ep);
+}
+
+void iser_conn_get(struct iser_conn *ib_conn)
+{
+	atomic_inc(&ib_conn->refcount);
+}
+
+void iser_conn_put(struct iser_conn *ib_conn)
+{
+	if (atomic_dec_and_test(&ib_conn->refcount))
+		iser_conn_release(ib_conn);
+}
+
+/**
+ * triggers start of the disconnect procedures and wait for them to be done
+ */
+void iser_conn_terminate(struct iser_conn *ib_conn)
+{
+	int err = 0;
+
+	/* change the ib conn state only if the conn is UP, however always call
+	 * rdma_disconnect since this is the only way to cause the CMA to change
+	 * the QP state to ERROR
+	 */
+
+	iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP, ISER_CONN_TERMINATING);
+	err = rdma_disconnect(ib_conn->cma_id);
+	if (err)
+		iser_err("Failed to disconnect, conn: 0x%p err %d\n",
+			 ib_conn,err);
+
+	wait_event_interruptible(ib_conn->wait,
+				 ib_conn->state == ISER_CONN_DOWN);
+
+	iser_conn_put(ib_conn);
+}
+
+static void iser_connect_error(struct rdma_cm_id *cma_id)
+{
+	struct iser_conn *ib_conn;
+	ib_conn = (struct iser_conn *)cma_id->context;
+
+	ib_conn->state = ISER_CONN_DOWN;
+	wake_up_interruptible(&ib_conn->wait);
+}
+
+static void iser_addr_handler(struct rdma_cm_id *cma_id)
+{
+	struct iser_device *device;
+	struct iser_conn   *ib_conn;
+	int    ret;
+
+	device = iser_device_find_by_ib_device(cma_id);
+	if (!device) {
+		iser_err("device lookup/creation failed\n");
+		iser_connect_error(cma_id);
+		return;
+	}
+
+	ib_conn = (struct iser_conn *)cma_id->context;
+	ib_conn->device = device;
+
+	ret = rdma_resolve_route(cma_id, 1000);
+	if (ret) {
+		iser_err("resolve route failed: %d\n", ret);
+		iser_connect_error(cma_id);
+	}
+}
+
+static void iser_route_handler(struct rdma_cm_id *cma_id)
+{
+	struct rdma_conn_param conn_param;
+	int    ret;
+
+	ret = iser_create_ib_conn_res((struct iser_conn *)cma_id->context);
+	if (ret)
+		goto failure;
+
+	memset(&conn_param, 0, sizeof conn_param);
+	conn_param.responder_resources = 4;
+	conn_param.initiator_depth     = 1;
+	conn_param.retry_count	       = 7;
+	conn_param.rnr_retry_count     = 6;
+
+	ret = rdma_connect(cma_id, &conn_param);
+	if (ret) {
+		iser_err("failure connecting: %d\n", ret);
+		goto failure;
+	}
+
+	return;
+failure:
+	iser_connect_error(cma_id);
+}
+
+static void iser_connected_handler(struct rdma_cm_id *cma_id)
+{
+	struct iser_conn *ib_conn;
+
+	ib_conn = (struct iser_conn *)cma_id->context;
+	ib_conn->state = ISER_CONN_UP;
+	wake_up_interruptible(&ib_conn->wait);
+}
+
+static void iser_disconnected_handler(struct rdma_cm_id *cma_id)
+{
+	struct iser_conn *ib_conn;
+
+	ib_conn = (struct iser_conn *)cma_id->context;
+	ib_conn->disc_evt_flag = 1;
+
+	/* getting here when the state is UP means that the conn is being *
+	 * terminated asynchronously from the iSCSI layer's perspective.  */
+	if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
+				      ISER_CONN_TERMINATING))
+		iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
+				   ISCSI_ERR_CONN_FAILED);
+
+	/* Complete the termination process if no posts are pending */
+	if ((atomic_read(&ib_conn->post_recv_buf_count) == 0) &&
+	    (atomic_read(&ib_conn->post_send_buf_count) == 0)) {
+		ib_conn->state = ISER_CONN_DOWN;
+		wake_up_interruptible(&ib_conn->wait);
+	}
+}
+
+static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
+{
+	int ret = 0;
+
+	iser_err("event %d conn %p id %p\n",event->event,cma_id->context,cma_id);
+
+	switch (event->event) {
+	case RDMA_CM_EVENT_ADDR_RESOLVED:
+		iser_addr_handler(cma_id);
+		break;
+	case RDMA_CM_EVENT_ROUTE_RESOLVED:
+		iser_route_handler(cma_id);
+		break;
+	case RDMA_CM_EVENT_ESTABLISHED:
+		iser_connected_handler(cma_id);
+		break;
+	case RDMA_CM_EVENT_ADDR_ERROR:
+	case RDMA_CM_EVENT_ROUTE_ERROR:
+	case RDMA_CM_EVENT_CONNECT_ERROR:
+	case RDMA_CM_EVENT_UNREACHABLE:
+	case RDMA_CM_EVENT_REJECTED:
+		iser_err("event: %d, error: %d\n", event->event, event->status);
+		iser_connect_error(cma_id);
+		break;
+	case RDMA_CM_EVENT_DISCONNECTED:
+	case RDMA_CM_EVENT_DEVICE_REMOVAL:
+	case RDMA_CM_EVENT_ADDR_CHANGE:
+		iser_disconnected_handler(cma_id);
+		break;
+	default:
+		iser_err("Unexpected RDMA CM event (%d)\n", event->event);
+		break;
+	}
+	return ret;
+}
+
+void iser_conn_init(struct iser_conn *ib_conn)
+{
+	ib_conn->state = ISER_CONN_INIT;
+	init_waitqueue_head(&ib_conn->wait);
+	atomic_set(&ib_conn->post_recv_buf_count, 0);
+	atomic_set(&ib_conn->post_send_buf_count, 0);
+	atomic_set(&ib_conn->unexpected_pdu_count, 0);
+	atomic_set(&ib_conn->refcount, 1);
+	INIT_LIST_HEAD(&ib_conn->conn_list);
+	spin_lock_init(&ib_conn->lock);
+}
+
+ /**
+ * starts the process of connecting to the target
+ * sleeps untill the connection is established or rejected
+ */
+int iser_connect(struct iser_conn   *ib_conn,
+		 struct sockaddr_in *src_addr,
+		 struct sockaddr_in *dst_addr,
+		 int                 non_blocking)
+{
+	struct sockaddr *src, *dst;
+	int err = 0;
+
+	sprintf(ib_conn->name, "%pI4:%d",
+		&dst_addr->sin_addr.s_addr, dst_addr->sin_port);
+
+	/* the device is known only --after-- address resolution */
+	ib_conn->device = NULL;
+
+	iser_err("connecting to: %pI4, port 0x%x\n",
+		 &dst_addr->sin_addr, dst_addr->sin_port);
+
+	ib_conn->state = ISER_CONN_PENDING;
+
+	ib_conn->cma_id = rdma_create_id(iser_cma_handler,
+					     (void *)ib_conn,
+					     RDMA_PS_TCP);
+	if (IS_ERR(ib_conn->cma_id)) {
+		err = PTR_ERR(ib_conn->cma_id);
+		iser_err("rdma_create_id failed: %d\n", err);
+		goto id_failure;
+	}
+
+	src = (struct sockaddr *)src_addr;
+	dst = (struct sockaddr *)dst_addr;
+	err = rdma_resolve_addr(ib_conn->cma_id, src, dst, 1000);
+	if (err) {
+		iser_err("rdma_resolve_addr failed: %d\n", err);
+		goto addr_failure;
+	}
+
+	if (!non_blocking) {
+		wait_event_interruptible(ib_conn->wait,
+					 (ib_conn->state != ISER_CONN_PENDING));
+
+		if (ib_conn->state != ISER_CONN_UP) {
+			err =  -EIO;
+			goto connect_failure;
+		}
+	}
+
+	mutex_lock(&ig.connlist_mutex);
+	list_add(&ib_conn->conn_list, &ig.connlist);
+	mutex_unlock(&ig.connlist_mutex);
+	return 0;
+
+id_failure:
+	ib_conn->cma_id = NULL;
+addr_failure:
+	ib_conn->state = ISER_CONN_DOWN;
+connect_failure:
+	iser_conn_release(ib_conn);
+	return err;
+}
+
+/**
+ * iser_reg_page_vec - Register physical memory
+ *
+ * returns: 0 on success, errno code on failure
+ */
+int iser_reg_page_vec(struct iser_conn     *ib_conn,
+		      struct iser_page_vec *page_vec,
+		      struct iser_mem_reg  *mem_reg)
+{
+	struct ib_pool_fmr *mem;
+	u64		   io_addr;
+	u64		   *page_list;
+	int		   status;
+
+	page_list = page_vec->pages;
+	io_addr	  = page_list[0];
+
+	mem  = ib_fmr_pool_map_phys(ib_conn->fmr_pool,
+				    page_list,
+				    page_vec->length,
+				    io_addr);
+
+	if (IS_ERR(mem)) {
+		status = (int)PTR_ERR(mem);
+		iser_err("ib_fmr_pool_map_phys failed: %d\n", status);
+		return status;
+	}
+
+	mem_reg->lkey  = mem->fmr->lkey;
+	mem_reg->rkey  = mem->fmr->rkey;
+	mem_reg->len   = page_vec->length * SIZE_4K;
+	mem_reg->va    = io_addr;
+	mem_reg->is_fmr = 1;
+	mem_reg->mem_h = (void *)mem;
+
+	mem_reg->va   += page_vec->offset;
+	mem_reg->len   = page_vec->data_size;
+
+	iser_dbg("PHYSICAL Mem.register, [PHYS p_array: 0x%p, sz: %d, "
+		 "entry[0]: (0x%08lx,%ld)] -> "
+		 "[lkey: 0x%08X mem_h: 0x%p va: 0x%08lX sz: %ld]\n",
+		 page_vec, page_vec->length,
+		 (unsigned long)page_vec->pages[0],
+		 (unsigned long)page_vec->data_size,
+		 (unsigned int)mem_reg->lkey, mem_reg->mem_h,
+		 (unsigned long)mem_reg->va, (unsigned long)mem_reg->len);
+	return 0;
+}
+
+/**
+ * Unregister (previosuly registered) memory.
+ */
+void iser_unreg_mem(struct iser_mem_reg *reg)
+{
+	int ret;
+
+	iser_dbg("PHYSICAL Mem.Unregister mem_h %p\n",reg->mem_h);
+
+	ret = ib_fmr_pool_unmap((struct ib_pool_fmr *)reg->mem_h);
+	if (ret)
+		iser_err("ib_fmr_pool_unmap failed %d\n", ret);
+
+	reg->mem_h = NULL;
+}
+
+/**
+ * iser_dto_to_iov - builds IOV from a dto descriptor
+ */
+static void iser_dto_to_iov(struct iser_dto *dto, struct ib_sge *iov, int iov_len)
+{
+	int		     i;
+	struct ib_sge	     *sge;
+	struct iser_regd_buf *regd_buf;
+
+	if (dto->regd_vector_len > iov_len) {
+		iser_err("iov size %d too small for posting dto of len %d\n",
+			 iov_len, dto->regd_vector_len);
+		BUG();
+	}
+
+	for (i = 0; i < dto->regd_vector_len; i++) {
+		sge	    = &iov[i];
+		regd_buf  = dto->regd[i];
+
+		sge->addr   = regd_buf->reg.va;
+		sge->length = regd_buf->reg.len;
+		sge->lkey   = regd_buf->reg.lkey;
+
+		if (dto->used_sz[i] > 0)  /* Adjust size */
+			sge->length = dto->used_sz[i];
+
+		/* offset and length should not exceed the regd buf length */
+		if (sge->length + dto->offset[i] > regd_buf->reg.len) {
+			iser_err("Used len:%ld + offset:%d, exceed reg.buf.len:"
+				 "%ld in dto:0x%p [%d], va:0x%08lX\n",
+				 (unsigned long)sge->length, dto->offset[i],
+				 (unsigned long)regd_buf->reg.len, dto, i,
+				 (unsigned long)sge->addr);
+			BUG();
+		}
+
+		sge->addr += dto->offset[i]; /* Adjust offset */
+	}
+}
+
+/**
+ * iser_post_recv - Posts a receive buffer.
+ *
+ * returns 0 on success, -1 on failure
+ */
+int iser_post_recv(struct iser_desc *rx_desc)
+{
+	int		  ib_ret, ret_val = 0;
+	struct ib_recv_wr recv_wr, *recv_wr_failed;
+	struct ib_sge	  iov[2];
+	struct iser_conn  *ib_conn;
+	struct iser_dto   *recv_dto = &rx_desc->dto;
+
+	/* Retrieve conn */
+	ib_conn = recv_dto->ib_conn;
+
+	iser_dto_to_iov(recv_dto, iov, 2);
+
+	recv_wr.next	= NULL;
+	recv_wr.sg_list = iov;
+	recv_wr.num_sge = recv_dto->regd_vector_len;
+	recv_wr.wr_id	= (unsigned long)rx_desc;
+
+	atomic_inc(&ib_conn->post_recv_buf_count);
+	ib_ret	= ib_post_recv(ib_conn->qp, &recv_wr, &recv_wr_failed);
+	if (ib_ret) {
+		iser_err("ib_post_recv failed ret=%d\n", ib_ret);
+		atomic_dec(&ib_conn->post_recv_buf_count);
+		ret_val = -1;
+	}
+
+	return ret_val;
+}
+
+/**
+ * iser_start_send - Initiate a Send DTO operation
+ *
+ * returns 0 on success, -1 on failure
+ */
+int iser_post_send(struct iser_desc *tx_desc)
+{
+	int		  ib_ret, ret_val = 0;
+	struct ib_send_wr send_wr, *send_wr_failed;
+	struct ib_sge	  iov[MAX_REGD_BUF_VECTOR_LEN];
+	struct iser_conn  *ib_conn;
+	struct iser_dto   *dto = &tx_desc->dto;
+
+	ib_conn = dto->ib_conn;
+
+	iser_dto_to_iov(dto, iov, MAX_REGD_BUF_VECTOR_LEN);
+
+	send_wr.next	   = NULL;
+	send_wr.wr_id	   = (unsigned long)tx_desc;
+	send_wr.sg_list	   = iov;
+	send_wr.num_sge	   = dto->regd_vector_len;
+	send_wr.opcode	   = IB_WR_SEND;
+	send_wr.send_flags = dto->notify_enable ? IB_SEND_SIGNALED : 0;
+
+	atomic_inc(&ib_conn->post_send_buf_count);
+
+	ib_ret = ib_post_send(ib_conn->qp, &send_wr, &send_wr_failed);
+	if (ib_ret) {
+		iser_err("Failed to start SEND DTO, dto: 0x%p, IOV len: %d\n",
+			 dto, dto->regd_vector_len);
+		iser_err("ib_post_send failed, ret:%d\n", ib_ret);
+		atomic_dec(&ib_conn->post_send_buf_count);
+		ret_val = -1;
+	}
+
+	return ret_val;
+}
+
+static void iser_handle_comp_error(struct iser_desc *desc)
+{
+	struct iser_dto  *dto     = &desc->dto;
+	struct iser_conn *ib_conn = dto->ib_conn;
+
+	iser_dto_buffs_release(dto);
+
+	if (desc->type == ISCSI_RX) {
+		kfree(desc->data);
+		kmem_cache_free(ig.desc_cache, desc);
+		atomic_dec(&ib_conn->post_recv_buf_count);
+	} else { /* type is TX control/command/dataout */
+		if (desc->type == ISCSI_TX_DATAOUT)
+			kmem_cache_free(ig.desc_cache, desc);
+		atomic_dec(&ib_conn->post_send_buf_count);
+	}
+
+	if (atomic_read(&ib_conn->post_recv_buf_count) == 0 &&
+	    atomic_read(&ib_conn->post_send_buf_count) == 0) {
+		/* getting here when the state is UP means that the conn is *
+		 * being terminated asynchronously from the iSCSI layer's   *
+		 * perspective.                                             */
+		if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
+		    ISER_CONN_TERMINATING))
+			iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
+					   ISCSI_ERR_CONN_FAILED);
+
+		/* complete the termination process if disconnect event was delivered *
+		 * note there are no more non completed posts to the QP               */
+		if (ib_conn->disc_evt_flag) {
+			ib_conn->state = ISER_CONN_DOWN;
+			wake_up_interruptible(&ib_conn->wait);
+		}
+	}
+}
+
+static void iser_cq_tasklet_fn(unsigned long data)
+{
+	 struct iser_device  *device = (struct iser_device *)data;
+	 struct ib_cq	     *cq = device->cq;
+	 struct ib_wc	     wc;
+	 struct iser_desc    *desc;
+	 unsigned long	     xfer_len;
+
+	while (ib_poll_cq(cq, 1, &wc) == 1) {
+		desc	 = (struct iser_desc *) (unsigned long) wc.wr_id;
+		BUG_ON(desc == NULL);
+
+		if (wc.status == IB_WC_SUCCESS) {
+			if (desc->type == ISCSI_RX) {
+				xfer_len = (unsigned long)wc.byte_len;
+				iser_rcv_completion(desc, xfer_len);
+			} else /* type == ISCSI_TX_CONTROL/SCSI_CMD/DOUT */
+				iser_snd_completion(desc);
+		} else {
+			iser_err("comp w. error op %d status %d\n",desc->type,wc.status);
+			iser_handle_comp_error(desc);
+		}
+	}
+	/* #warning "it is assumed here that arming CQ only once its empty" *
+	 * " would not cause interrupts to be missed"                       */
+	ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
+}
+
+static void iser_cq_callback(struct ib_cq *cq, void *cq_context)
+{
+	struct iser_device  *device = (struct iser_device *)cq_context;
+
+	tasklet_schedule(&device->cq_tasklet);
+}

kernel/drivers/infiniband/ulp/srp/Kbuild

@@ -0,0 +1 @@
+obj-$(CONFIG_INFINIBAND_SRP)			+= ib_srp.o

kernel/drivers/infiniband/ulp/srp/Kconfig

@@ -0,0 +1,12 @@
+config INFINIBAND_SRP
+	tristate "InfiniBand SCSI RDMA Protocol"
+	depends on SCSI
+	select SCSI_SRP_ATTRS
+	---help---
+	  Support for the SCSI RDMA Protocol over InfiniBand.  This
+	  allows you to access storage devices that speak SRP over
+	  InfiniBand.
+
+	  The SRP protocol is defined by the INCITS T10 technical
+	  committee.  See <http://www.t10.org/>.
+

kernel/drivers/infiniband/ulp/srp/ib_srp.h

@@ -0,0 +1,171 @@
+/*
+ * Copyright (c) 2005 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.
+ */
+
+#ifndef IB_SRP_H
+#define IB_SRP_H
+
+#include <linux/types.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/scatterlist.h>
+
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_cmnd.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_sa.h>
+#include <rdma/ib_cm.h>
+#include <rdma/ib_fmr_pool.h>
+
+enum {
+	SRP_PATH_REC_TIMEOUT_MS	= 1000,
+	SRP_ABORT_TIMEOUT_MS	= 5000,
+
+	SRP_PORT_REDIRECT	= 1,
+	SRP_DLID_REDIRECT	= 2,
+	SRP_STALE_CONN		= 3,
+
+	SRP_MAX_LUN		= 512,
+	SRP_DEF_SG_TABLESIZE	= 12,
+
+	SRP_RQ_SHIFT    	= 6,
+	SRP_RQ_SIZE		= 1 << SRP_RQ_SHIFT,
+	SRP_SQ_SIZE		= SRP_RQ_SIZE - 1,
+	SRP_CQ_SIZE		= SRP_SQ_SIZE + SRP_RQ_SIZE,
+
+	SRP_TAG_TSK_MGMT	= 1 << (SRP_RQ_SHIFT + 1),
+
+	SRP_FMR_SIZE		= 256,
+	SRP_FMR_POOL_SIZE	= 1024,
+	SRP_FMR_DIRTY_SIZE	= SRP_FMR_POOL_SIZE / 4
+};
+
+#define SRP_OP_RECV		(1 << 31)
+
+enum srp_target_state {
+	SRP_TARGET_LIVE,
+	SRP_TARGET_CONNECTING,
+	SRP_TARGET_DEAD,
+	SRP_TARGET_REMOVED
+};
+
+enum srp_request_type {
+	SRP_REQ_NORMAL,
+	SRP_REQ_TASK_MGMT,
+};
+
+struct srp_device {
+	struct list_head	dev_list;
+	struct ib_device       *dev;
+	struct ib_pd	       *pd;
+	struct ib_mr	       *mr;
+	struct ib_fmr_pool     *fmr_pool;
+	int			fmr_page_shift;
+	int			fmr_page_size;
+	u64			fmr_page_mask;
+};
+
+struct srp_host {
+	struct srp_device      *srp_dev;
+	u8			port;
+	struct device		dev;
+	struct list_head	target_list;
+	spinlock_t		target_lock;
+	struct completion	released;
+	struct list_head	list;
+};
+
+struct srp_request {
+	struct list_head	list;
+	struct scsi_cmnd       *scmnd;
+	struct srp_iu	       *cmd;
+	struct srp_iu	       *tsk_mgmt;
+	struct ib_pool_fmr     *fmr;
+	struct completion	done;
+	short			index;
+	u8			cmd_done;
+	u8			tsk_status;
+};
+
+struct srp_target_port {
+	__be64			id_ext;
+	__be64			ioc_guid;
+	__be64			service_id;
+	__be64			initiator_ext;
+	u16			io_class;
+	struct srp_host	       *srp_host;
+	struct Scsi_Host       *scsi_host;
+	char			target_name[32];
+	unsigned int		scsi_id;
+
+	struct ib_sa_path_rec	path;
+	__be16			orig_dgid[8];
+	struct ib_sa_query     *path_query;
+	int			path_query_id;
+
+	struct ib_cm_id	       *cm_id;
+	struct ib_cq	       *cq;
+	struct ib_qp	       *qp;
+
+	int			max_ti_iu_len;
+	s32			req_lim;
+
+	int			zero_req_lim;
+
+	unsigned		rx_head;
+	struct srp_iu	       *rx_ring[SRP_RQ_SIZE];
+
+	unsigned		tx_head;
+	unsigned		tx_tail;
+	struct srp_iu	       *tx_ring[SRP_SQ_SIZE + 1];
+
+	struct list_head	free_reqs;
+	struct list_head	req_queue;
+	struct srp_request	req_ring[SRP_SQ_SIZE];
+
+	struct work_struct	work;
+
+	struct list_head	list;
+	struct completion	done;
+	int			status;
+	enum srp_target_state	state;
+	int			qp_in_error;
+};
+
+struct srp_iu {
+	u64			dma;
+	void		       *buf;
+	size_t			size;
+	enum dma_data_direction	direction;
+};
+
+#endif /* IB_SRP_H */