Initial commit; kernel source import

drivers/infiniband/hw/mlx4/Kconfig

@@ -0,0 +1,10 @@
+config MLX4_INFINIBAND
+	tristate "Mellanox ConnectX HCA support"
+	depends on NETDEVICES && ETHERNET && PCI
+	select NET_VENDOR_MELLANOX
+	select MLX4_CORE
+	---help---
+	  This driver provides low-level InfiniBand support for
+	  Mellanox ConnectX PCI Express host channel adapters (HCAs).
+	  This is required to use InfiniBand protocols such as
+	  IP-over-IB or SRP with these devices.

drivers/infiniband/hw/mlx4/Makefile

@@ -0,0 +1,3 @@
+obj-$(CONFIG_MLX4_INFINIBAND)	+= mlx4_ib.o
+
+mlx4_ib-y :=	ah.o cq.o doorbell.o mad.o main.o mr.o qp.o srq.o mcg.o cm.o alias_GUID.o sysfs.o

drivers/infiniband/hw/mlx4/ah.c

@@ -0,0 +1,199 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, 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 <rdma/ib_addr.h>
+#include <rdma/ib_cache.h>
+
+#include <linux/slab.h>
+#include <linux/inet.h>
+#include <linux/string.h>
+
+#include "mlx4_ib.h"
+
+int mlx4_ib_resolve_grh(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah_attr,
+			u8 *mac, int *is_mcast, u8 port)
+{
+	struct in6_addr in6;
+
+	*is_mcast = 0;
+
+	memcpy(&in6, ah_attr->grh.dgid.raw, sizeof in6);
+	if (rdma_link_local_addr(&in6))
+		rdma_get_ll_mac(&in6, mac);
+	else if (rdma_is_multicast_addr(&in6)) {
+		rdma_get_mcast_mac(&in6, mac);
+		*is_mcast = 1;
+	} else
+		return -EINVAL;
+
+	return 0;
+}
+
+static struct ib_ah *create_ib_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr,
+				  struct mlx4_ib_ah *ah)
+{
+	struct mlx4_dev *dev = to_mdev(pd->device)->dev;
+
+	ah->av.ib.port_pd = cpu_to_be32(to_mpd(pd)->pdn | (ah_attr->port_num << 24));
+	ah->av.ib.g_slid  = ah_attr->src_path_bits;
+	if (ah_attr->ah_flags & IB_AH_GRH) {
+		ah->av.ib.g_slid   |= 0x80;
+		ah->av.ib.gid_index = ah_attr->grh.sgid_index;
+		ah->av.ib.hop_limit = ah_attr->grh.hop_limit;
+		ah->av.ib.sl_tclass_flowlabel |=
+			cpu_to_be32((ah_attr->grh.traffic_class << 20) |
+				    ah_attr->grh.flow_label);
+		memcpy(ah->av.ib.dgid, ah_attr->grh.dgid.raw, 16);
+	}
+
+	ah->av.ib.dlid    = cpu_to_be16(ah_attr->dlid);
+	if (ah_attr->static_rate) {
+		ah->av.ib.stat_rate = ah_attr->static_rate + MLX4_STAT_RATE_OFFSET;
+		while (ah->av.ib.stat_rate > IB_RATE_2_5_GBPS + MLX4_STAT_RATE_OFFSET &&
+		       !(1 << ah->av.ib.stat_rate & dev->caps.stat_rate_support))
+			--ah->av.ib.stat_rate;
+	}
+	ah->av.ib.sl_tclass_flowlabel = cpu_to_be32(ah_attr->sl << 28);
+
+	return &ah->ibah;
+}
+
+static struct ib_ah *create_iboe_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr,
+				    struct mlx4_ib_ah *ah)
+{
+	struct mlx4_ib_dev *ibdev = to_mdev(pd->device);
+	struct mlx4_dev *dev = ibdev->dev;
+	union ib_gid sgid;
+	u8 mac[6];
+	int err;
+	int is_mcast;
+	u16 vlan_tag;
+
+	err = mlx4_ib_resolve_grh(ibdev, ah_attr, mac, &is_mcast, ah_attr->port_num);
+	if (err)
+		return ERR_PTR(err);
+
+	memcpy(ah->av.eth.mac, mac, 6);
+	err = ib_get_cached_gid(pd->device, ah_attr->port_num, ah_attr->grh.sgid_index, &sgid);
+	if (err)
+		return ERR_PTR(err);
+	vlan_tag = rdma_get_vlan_id(&sgid);
+	if (vlan_tag < 0x1000)
+		vlan_tag |= (ah_attr->sl & 7) << 13;
+	ah->av.eth.port_pd = cpu_to_be32(to_mpd(pd)->pdn | (ah_attr->port_num << 24));
+	ah->av.eth.gid_index = ah_attr->grh.sgid_index;
+	ah->av.eth.vlan = cpu_to_be16(vlan_tag);
+	if (ah_attr->static_rate) {
+		ah->av.eth.stat_rate = ah_attr->static_rate + MLX4_STAT_RATE_OFFSET;
+		while (ah->av.eth.stat_rate > IB_RATE_2_5_GBPS + MLX4_STAT_RATE_OFFSET &&
+		       !(1 << ah->av.eth.stat_rate & dev->caps.stat_rate_support))
+			--ah->av.eth.stat_rate;
+	}
+
+	/*
+	 * HW requires multicast LID so we just choose one.
+	 */
+	if (is_mcast)
+		ah->av.ib.dlid = cpu_to_be16(0xc000);
+
+	memcpy(ah->av.eth.dgid, ah_attr->grh.dgid.raw, 16);
+	ah->av.eth.sl_tclass_flowlabel = cpu_to_be32(ah_attr->sl << 29);
+
+	return &ah->ibah;
+}
+
+struct ib_ah *mlx4_ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr)
+{
+	struct mlx4_ib_ah *ah;
+	struct ib_ah *ret;
+
+	ah = kzalloc(sizeof *ah, GFP_ATOMIC);
+	if (!ah)
+		return ERR_PTR(-ENOMEM);
+
+	if (rdma_port_get_link_layer(pd->device, ah_attr->port_num) == IB_LINK_LAYER_ETHERNET) {
+		if (!(ah_attr->ah_flags & IB_AH_GRH)) {
+			ret = ERR_PTR(-EINVAL);
+		} else {
+			/*
+			 * TBD: need to handle the case when we get
+			 * called in an atomic context and there we
+			 * might sleep.  We don't expect this
+			 * currently since we're working with link
+			 * local addresses which we can translate
+			 * without going to sleep.
+			 */
+			ret = create_iboe_ah(pd, ah_attr, ah);
+		}
+
+		if (IS_ERR(ret))
+			kfree(ah);
+
+		return ret;
+	} else
+		return create_ib_ah(pd, ah_attr, ah); /* never fails */
+}
+
+int mlx4_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr)
+{
+	struct mlx4_ib_ah *ah = to_mah(ibah);
+	enum rdma_link_layer ll;
+
+	memset(ah_attr, 0, sizeof *ah_attr);
+	ah_attr->sl = be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 28;
+	ah_attr->port_num = be32_to_cpu(ah->av.ib.port_pd) >> 24;
+	ll = rdma_port_get_link_layer(ibah->device, ah_attr->port_num);
+	ah_attr->dlid = ll == IB_LINK_LAYER_INFINIBAND ? be16_to_cpu(ah->av.ib.dlid) : 0;
+	if (ah->av.ib.stat_rate)
+		ah_attr->static_rate = ah->av.ib.stat_rate - MLX4_STAT_RATE_OFFSET;
+	ah_attr->src_path_bits = ah->av.ib.g_slid & 0x7F;
+
+	if (mlx4_ib_ah_grh_present(ah)) {
+		ah_attr->ah_flags = IB_AH_GRH;
+
+		ah_attr->grh.traffic_class =
+			be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 20;
+		ah_attr->grh.flow_label =
+			be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) & 0xfffff;
+		ah_attr->grh.hop_limit  = ah->av.ib.hop_limit;
+		ah_attr->grh.sgid_index = ah->av.ib.gid_index;
+		memcpy(ah_attr->grh.dgid.raw, ah->av.ib.dgid, 16);
+	}
+
+	return 0;
+}
+
+int mlx4_ib_destroy_ah(struct ib_ah *ah)
+{
+	kfree(to_mah(ah));
+	return 0;
+}

drivers/infiniband/hw/mlx4/alias_GUID.c

@@ -0,0 +1,688 @@
+/*
+ * Copyright (c) 2012 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.
+ */
+ /***********************************************************/
+/*This file support the handling of the Alias GUID feature. */
+/***********************************************************/
+#include <rdma/ib_mad.h>
+#include <rdma/ib_smi.h>
+#include <rdma/ib_cache.h>
+#include <rdma/ib_sa.h>
+#include <rdma/ib_pack.h>
+#include <linux/mlx4/cmd.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <rdma/ib_user_verbs.h>
+#include <linux/delay.h>
+#include "mlx4_ib.h"
+
+/*
+The driver keeps the current state of all guids, as they are in the HW.
+Whenever we receive an smp mad GUIDInfo record, the data will be cached.
+*/
+
+struct mlx4_alias_guid_work_context {
+	u8 port;
+	struct mlx4_ib_dev     *dev ;
+	struct ib_sa_query     *sa_query;
+	struct completion	done;
+	int			query_id;
+	struct list_head	list;
+	int			block_num;
+};
+
+struct mlx4_next_alias_guid_work {
+	u8 port;
+	u8 block_num;
+	struct mlx4_sriov_alias_guid_info_rec_det rec_det;
+};
+
+
+void mlx4_ib_update_cache_on_guid_change(struct mlx4_ib_dev *dev, int block_num,
+					 u8 port_num, u8 *p_data)
+{
+	int i;
+	u64 guid_indexes;
+	int slave_id;
+	int port_index = port_num - 1;
+
+	if (!mlx4_is_master(dev->dev))
+		return;
+
+	guid_indexes = be64_to_cpu((__force __be64) dev->sriov.alias_guid.
+				   ports_guid[port_num - 1].
+				   all_rec_per_port[block_num].guid_indexes);
+	pr_debug("port: %d, guid_indexes: 0x%llx\n", port_num, guid_indexes);
+
+	for (i = 0; i < NUM_ALIAS_GUID_IN_REC; i++) {
+		/* The location of the specific index starts from bit number 4
+		 * until bit num 11 */
+		if (test_bit(i + 4, (unsigned long *)&guid_indexes)) {
+			slave_id = (block_num * NUM_ALIAS_GUID_IN_REC) + i ;
+			if (slave_id >= dev->dev->num_slaves) {
+				pr_debug("The last slave: %d\n", slave_id);
+				return;
+			}
+
+			/* cache the guid: */
+			memcpy(&dev->sriov.demux[port_index].guid_cache[slave_id],
+			       &p_data[i * GUID_REC_SIZE],
+			       GUID_REC_SIZE);
+		} else
+			pr_debug("Guid number: %d in block: %d"
+				 " was not updated\n", i, block_num);
+	}
+}
+
+static __be64 get_cached_alias_guid(struct mlx4_ib_dev *dev, int port, int index)
+{
+	if (index >= NUM_ALIAS_GUID_PER_PORT) {
+		pr_err("%s: ERROR: asked for index:%d\n", __func__, index);
+		return (__force __be64) -1;
+	}
+	return *(__be64 *)&dev->sriov.demux[port - 1].guid_cache[index];
+}
+
+
+ib_sa_comp_mask mlx4_ib_get_aguid_comp_mask_from_ix(int index)
+{
+	return IB_SA_COMP_MASK(4 + index);
+}
+
+/*
+ * Whenever new GUID is set/unset (guid table change) create event and
+ * notify the relevant slave (master also should be notified).
+ * If the GUID value is not as we have in the cache the slave will not be
+ * updated; in this case it waits for the smp_snoop or the port management
+ * event to call the function and to update the slave.
+ * block_number - the index of the block (16 blocks available)
+ * port_number - 1 or 2
+ */
+void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev,
+					  int block_num, u8 port_num,
+					  u8 *p_data)
+{
+	int i;
+	u64 guid_indexes;
+	int slave_id;
+	enum slave_port_state new_state;
+	enum slave_port_state prev_state;
+	__be64 tmp_cur_ag, form_cache_ag;
+	enum slave_port_gen_event gen_event;
+
+	if (!mlx4_is_master(dev->dev))
+		return;
+
+	guid_indexes = be64_to_cpu((__force __be64) dev->sriov.alias_guid.
+				   ports_guid[port_num - 1].
+				   all_rec_per_port[block_num].guid_indexes);
+	pr_debug("port: %d, guid_indexes: 0x%llx\n", port_num, guid_indexes);
+
+	/*calculate the slaves and notify them*/
+	for (i = 0; i < NUM_ALIAS_GUID_IN_REC; i++) {
+		/* the location of the specific index runs from bits 4..11 */
+		if (!(test_bit(i + 4, (unsigned long *)&guid_indexes)))
+			continue;
+
+		slave_id = (block_num * NUM_ALIAS_GUID_IN_REC) + i ;
+		if (slave_id >= dev->dev->num_slaves)
+			return;
+		tmp_cur_ag = *(__be64 *)&p_data[i * GUID_REC_SIZE];
+		form_cache_ag = get_cached_alias_guid(dev, port_num,
+					(NUM_ALIAS_GUID_IN_REC * block_num) + i);
+		/*
+		 * Check if guid is not the same as in the cache,
+		 * If it is different, wait for the snoop_smp or the port mgmt
+		 * change event to update the slave on its port state change
+		 */
+		if (tmp_cur_ag != form_cache_ag)
+			continue;
+		mlx4_gen_guid_change_eqe(dev->dev, slave_id, port_num);
+
+		/*2 cases: Valid GUID, and Invalid Guid*/
+
+		if (tmp_cur_ag != MLX4_NOT_SET_GUID) { /*valid GUID*/
+			prev_state = mlx4_get_slave_port_state(dev->dev, slave_id, port_num);
+			new_state = set_and_calc_slave_port_state(dev->dev, slave_id, port_num,
+								  MLX4_PORT_STATE_IB_PORT_STATE_EVENT_GID_VALID,
+								  &gen_event);
+			pr_debug("slave: %d, port: %d prev_port_state: %d,"
+				 " new_port_state: %d, gen_event: %d\n",
+				 slave_id, port_num, prev_state, new_state, gen_event);
+			if (gen_event == SLAVE_PORT_GEN_EVENT_UP) {
+				pr_debug("sending PORT_UP event to slave: %d, port: %d\n",
+					 slave_id, port_num);
+				mlx4_gen_port_state_change_eqe(dev->dev, slave_id,
+							       port_num, MLX4_PORT_CHANGE_SUBTYPE_ACTIVE);
+			}
+		} else { /* request to invalidate GUID */
+			set_and_calc_slave_port_state(dev->dev, slave_id, port_num,
+						      MLX4_PORT_STATE_IB_EVENT_GID_INVALID,
+						      &gen_event);
+			pr_debug("sending PORT DOWN event to slave: %d, port: %d\n",
+				 slave_id, port_num);
+			mlx4_gen_port_state_change_eqe(dev->dev, slave_id, port_num,
+						       MLX4_PORT_CHANGE_SUBTYPE_DOWN);
+		}
+	}
+}
+
+static void aliasguid_query_handler(int status,
+				    struct ib_sa_guidinfo_rec *guid_rec,
+				    void *context)
+{
+	struct mlx4_ib_dev *dev;
+	struct mlx4_alias_guid_work_context *cb_ctx = context;
+	u8 port_index ;
+	int i;
+	struct mlx4_sriov_alias_guid_info_rec_det *rec;
+	unsigned long flags, flags1;
+
+	if (!context)
+		return;
+
+	dev = cb_ctx->dev;
+	port_index = cb_ctx->port - 1;
+	rec = &dev->sriov.alias_guid.ports_guid[port_index].
+		all_rec_per_port[cb_ctx->block_num];
+
+	if (status) {
+		rec->status = MLX4_GUID_INFO_STATUS_IDLE;
+		pr_debug("(port: %d) failed: status = %d\n",
+			 cb_ctx->port, status);
+		goto out;
+	}
+
+	if (guid_rec->block_num != cb_ctx->block_num) {
+		pr_err("block num mismatch: %d != %d\n",
+		       cb_ctx->block_num, guid_rec->block_num);
+		goto out;
+	}
+
+	pr_debug("lid/port: %d/%d, block_num: %d\n",
+		 be16_to_cpu(guid_rec->lid), cb_ctx->port,
+		 guid_rec->block_num);
+
+	rec = &dev->sriov.alias_guid.ports_guid[port_index].
+		all_rec_per_port[guid_rec->block_num];
+
+	rec->status = MLX4_GUID_INFO_STATUS_SET;
+	rec->method = MLX4_GUID_INFO_RECORD_SET;
+
+	for (i = 0 ; i < NUM_ALIAS_GUID_IN_REC; i++) {
+		__be64 tmp_cur_ag;
+		tmp_cur_ag = *(__be64 *)&guid_rec->guid_info_list[i * GUID_REC_SIZE];
+		/* check if the SM didn't assign one of the records.
+		 * if it didn't, if it was not sysadmin request:
+		 * ask the SM to give a new GUID, (instead of the driver request).
+		 */
+		if (tmp_cur_ag == MLX4_NOT_SET_GUID) {
+			mlx4_ib_warn(&dev->ib_dev, "%s:Record num %d in "
+				     "block_num: %d was declined by SM, "
+				     "ownership by %d (0 = driver, 1=sysAdmin,"
+				     " 2=None)\n", __func__, i,
+				     guid_rec->block_num, rec->ownership);
+			if (rec->ownership == MLX4_GUID_DRIVER_ASSIGN) {
+				/* if it is driver assign, asks for new GUID from SM*/
+				*(__be64 *)&rec->all_recs[i * GUID_REC_SIZE] =
+					MLX4_NOT_SET_GUID;
+
+				/* Mark the record as not assigned, and let it
+				 * be sent again in the next work sched.*/
+				rec->status = MLX4_GUID_INFO_STATUS_IDLE;
+				rec->guid_indexes |= mlx4_ib_get_aguid_comp_mask_from_ix(i);
+			}
+		} else {
+		       /* properly assigned record. */
+		       /* We save the GUID we just got from the SM in the
+			* admin_guid in order to be persistent, and in the
+			* request from the sm the process will ask for the same GUID */
+			if (rec->ownership == MLX4_GUID_SYSADMIN_ASSIGN &&
+			    tmp_cur_ag != *(__be64 *)&rec->all_recs[i * GUID_REC_SIZE]) {
+				/* the sysadmin assignment failed.*/
+				mlx4_ib_warn(&dev->ib_dev, "%s: Failed to set"
+					     " admin guid after SysAdmin "
+					     "configuration. "
+					     "Record num %d in block_num:%d "
+					     "was declined by SM, "
+					     "new val(0x%llx) was kept\n",
+					      __func__, i,
+					     guid_rec->block_num,
+					     be64_to_cpu(*(__be64 *) &
+							 rec->all_recs[i * GUID_REC_SIZE]));
+			} else {
+				memcpy(&rec->all_recs[i * GUID_REC_SIZE],
+				       &guid_rec->guid_info_list[i * GUID_REC_SIZE],
+				       GUID_REC_SIZE);
+			}
+		}
+	}
+	/*
+	The func is call here to close the cases when the
+	sm doesn't send smp, so in the sa response the driver
+	notifies the slave.
+	*/
+	mlx4_ib_notify_slaves_on_guid_change(dev, guid_rec->block_num,
+					     cb_ctx->port,
+					     guid_rec->guid_info_list);
+out:
+	spin_lock_irqsave(&dev->sriov.going_down_lock, flags);
+	spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	if (!dev->sriov.is_going_down)
+		queue_delayed_work(dev->sriov.alias_guid.ports_guid[port_index].wq,
+				   &dev->sriov.alias_guid.ports_guid[port_index].
+				   alias_guid_work, 0);
+	if (cb_ctx->sa_query) {
+		list_del(&cb_ctx->list);
+		kfree(cb_ctx);
+	} else
+		complete(&cb_ctx->done);
+	spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	spin_unlock_irqrestore(&dev->sriov.going_down_lock, flags);
+}
+
+static void invalidate_guid_record(struct mlx4_ib_dev *dev, u8 port, int index)
+{
+	int i;
+	u64 cur_admin_val;
+	ib_sa_comp_mask comp_mask = 0;
+
+	dev->sriov.alias_guid.ports_guid[port - 1].all_rec_per_port[index].status
+		= MLX4_GUID_INFO_STATUS_IDLE;
+	dev->sriov.alias_guid.ports_guid[port - 1].all_rec_per_port[index].method
+		= MLX4_GUID_INFO_RECORD_SET;
+
+	/* calculate the comp_mask for that record.*/
+	for (i = 0; i < NUM_ALIAS_GUID_IN_REC; i++) {
+		cur_admin_val =
+			*(u64 *)&dev->sriov.alias_guid.ports_guid[port - 1].
+			all_rec_per_port[index].all_recs[GUID_REC_SIZE * i];
+		/*
+		check the admin value: if it's for delete (~00LL) or
+		it is the first guid of the first record (hw guid) or
+		the records is not in ownership of the sysadmin and the sm doesn't
+		need to assign GUIDs, then don't put it up for assignment.
+		*/
+		if (MLX4_GUID_FOR_DELETE_VAL == cur_admin_val ||
+		    (!index && !i) ||
+		    MLX4_GUID_NONE_ASSIGN == dev->sriov.alias_guid.
+		    ports_guid[port - 1].all_rec_per_port[index].ownership)
+			continue;
+		comp_mask |= mlx4_ib_get_aguid_comp_mask_from_ix(i);
+	}
+	dev->sriov.alias_guid.ports_guid[port - 1].
+		all_rec_per_port[index].guid_indexes = comp_mask;
+}
+
+static int set_guid_rec(struct ib_device *ibdev,
+			u8 port, int index,
+			struct mlx4_sriov_alias_guid_info_rec_det *rec_det)
+{
+	int err;
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct ib_sa_guidinfo_rec guid_info_rec;
+	ib_sa_comp_mask comp_mask;
+	struct ib_port_attr attr;
+	struct mlx4_alias_guid_work_context *callback_context;
+	unsigned long resched_delay, flags, flags1;
+	struct list_head *head =
+		&dev->sriov.alias_guid.ports_guid[port - 1].cb_list;
+
+	err = __mlx4_ib_query_port(ibdev, port, &attr, 1);
+	if (err) {
+		pr_debug("mlx4_ib_query_port failed (err: %d), port: %d\n",
+			 err, port);
+		return err;
+	}
+	/*check the port was configured by the sm, otherwise no need to send */
+	if (attr.state != IB_PORT_ACTIVE) {
+		pr_debug("port %d not active...rescheduling\n", port);
+		resched_delay = 5 * HZ;
+		err = -EAGAIN;
+		goto new_schedule;
+	}
+
+	callback_context = kmalloc(sizeof *callback_context, GFP_KERNEL);
+	if (!callback_context) {
+		err = -ENOMEM;
+		resched_delay = HZ * 5;
+		goto new_schedule;
+	}
+	callback_context->port = port;
+	callback_context->dev = dev;
+	callback_context->block_num = index;
+
+	memset(&guid_info_rec, 0, sizeof (struct ib_sa_guidinfo_rec));
+
+	guid_info_rec.lid = cpu_to_be16(attr.lid);
+	guid_info_rec.block_num = index;
+
+	memcpy(guid_info_rec.guid_info_list, rec_det->all_recs,
+	       GUID_REC_SIZE * NUM_ALIAS_GUID_IN_REC);
+	comp_mask = IB_SA_GUIDINFO_REC_LID | IB_SA_GUIDINFO_REC_BLOCK_NUM |
+		rec_det->guid_indexes;
+
+	init_completion(&callback_context->done);
+	spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	list_add_tail(&callback_context->list, head);
+	spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+
+	callback_context->query_id =
+		ib_sa_guid_info_rec_query(dev->sriov.alias_guid.sa_client,
+					  ibdev, port, &guid_info_rec,
+					  comp_mask, rec_det->method, 1000,
+					  GFP_KERNEL, aliasguid_query_handler,
+					  callback_context,
+					  &callback_context->sa_query);
+	if (callback_context->query_id < 0) {
+		pr_debug("ib_sa_guid_info_rec_query failed, query_id: "
+			 "%d. will reschedule to the next 1 sec.\n",
+			 callback_context->query_id);
+		spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+		list_del(&callback_context->list);
+		kfree(callback_context);
+		spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+		resched_delay = 1 * HZ;
+		err = -EAGAIN;
+		goto new_schedule;
+	}
+	err = 0;
+	goto out;
+
+new_schedule:
+	spin_lock_irqsave(&dev->sriov.going_down_lock, flags);
+	spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	invalidate_guid_record(dev, port, index);
+	if (!dev->sriov.is_going_down) {
+		queue_delayed_work(dev->sriov.alias_guid.ports_guid[port - 1].wq,
+				   &dev->sriov.alias_guid.ports_guid[port - 1].alias_guid_work,
+				   resched_delay);
+	}
+	spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	spin_unlock_irqrestore(&dev->sriov.going_down_lock, flags);
+
+out:
+	return err;
+}
+
+void mlx4_ib_invalidate_all_guid_record(struct mlx4_ib_dev *dev, int port)
+{
+	int i;
+	unsigned long flags, flags1;
+
+	pr_debug("port %d\n", port);
+
+	spin_lock_irqsave(&dev->sriov.going_down_lock, flags);
+	spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	for (i = 0; i < NUM_ALIAS_GUID_REC_IN_PORT; i++)
+		invalidate_guid_record(dev, port, i);
+
+	if (mlx4_is_master(dev->dev) && !dev->sriov.is_going_down) {
+		/*
+		make sure no work waits in the queue, if the work is already
+		queued(not on the timer) the cancel will fail. That is not a problem
+		because we just want the work started.
+		*/
+		cancel_delayed_work(&dev->sriov.alias_guid.
+				      ports_guid[port - 1].alias_guid_work);
+		queue_delayed_work(dev->sriov.alias_guid.ports_guid[port - 1].wq,
+				   &dev->sriov.alias_guid.ports_guid[port - 1].alias_guid_work,
+				   0);
+	}
+	spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	spin_unlock_irqrestore(&dev->sriov.going_down_lock, flags);
+}
+
+/* The function returns the next record that was
+ * not configured (or failed to be configured) */
+static int get_next_record_to_update(struct mlx4_ib_dev *dev, u8 port,
+				     struct mlx4_next_alias_guid_work *rec)
+{
+	int j;
+	unsigned long flags;
+
+	for (j = 0; j < NUM_ALIAS_GUID_REC_IN_PORT; j++) {
+		spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags);
+		if (dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[j].status ==
+		    MLX4_GUID_INFO_STATUS_IDLE) {
+			memcpy(&rec->rec_det,
+			       &dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[j],
+			       sizeof (struct mlx4_sriov_alias_guid_info_rec_det));
+			rec->port = port;
+			rec->block_num = j;
+			dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[j].status =
+				MLX4_GUID_INFO_STATUS_PENDING;
+			spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags);
+			return 0;
+		}
+		spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags);
+	}
+	return -ENOENT;
+}
+
+static void set_administratively_guid_record(struct mlx4_ib_dev *dev, int port,
+					     int rec_index,
+					     struct mlx4_sriov_alias_guid_info_rec_det *rec_det)
+{
+	dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[rec_index].guid_indexes =
+		rec_det->guid_indexes;
+	memcpy(dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[rec_index].all_recs,
+	       rec_det->all_recs, NUM_ALIAS_GUID_IN_REC * GUID_REC_SIZE);
+	dev->sriov.alias_guid.ports_guid[port].all_rec_per_port[rec_index].status =
+		rec_det->status;
+}
+
+static void set_all_slaves_guids(struct mlx4_ib_dev *dev, int port)
+{
+	int j;
+	struct mlx4_sriov_alias_guid_info_rec_det rec_det ;
+
+	for (j = 0 ; j < NUM_ALIAS_GUID_REC_IN_PORT ; j++) {
+		memset(rec_det.all_recs, 0, NUM_ALIAS_GUID_IN_REC * GUID_REC_SIZE);
+		rec_det.guid_indexes = (!j ? 0 : IB_SA_GUIDINFO_REC_GID0) |
+			IB_SA_GUIDINFO_REC_GID1 | IB_SA_GUIDINFO_REC_GID2 |
+			IB_SA_GUIDINFO_REC_GID3 | IB_SA_GUIDINFO_REC_GID4 |
+			IB_SA_GUIDINFO_REC_GID5 | IB_SA_GUIDINFO_REC_GID6 |
+			IB_SA_GUIDINFO_REC_GID7;
+		rec_det.status = MLX4_GUID_INFO_STATUS_IDLE;
+		set_administratively_guid_record(dev, port, j, &rec_det);
+	}
+}
+
+static void alias_guid_work(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	int ret = 0;
+	struct mlx4_next_alias_guid_work *rec;
+	struct mlx4_sriov_alias_guid_port_rec_det *sriov_alias_port =
+		container_of(delay, struct mlx4_sriov_alias_guid_port_rec_det,
+			     alias_guid_work);
+	struct mlx4_sriov_alias_guid *sriov_alias_guid = sriov_alias_port->parent;
+	struct mlx4_ib_sriov *ib_sriov = container_of(sriov_alias_guid,
+						struct mlx4_ib_sriov,
+						alias_guid);
+	struct mlx4_ib_dev *dev = container_of(ib_sriov, struct mlx4_ib_dev, sriov);
+
+	rec = kzalloc(sizeof *rec, GFP_KERNEL);
+	if (!rec) {
+		pr_err("alias_guid_work: No Memory\n");
+		return;
+	}
+
+	pr_debug("starting [port: %d]...\n", sriov_alias_port->port + 1);
+	ret = get_next_record_to_update(dev, sriov_alias_port->port, rec);
+	if (ret) {
+		pr_debug("No more records to update.\n");
+		goto out;
+	}
+
+	set_guid_rec(&dev->ib_dev, rec->port + 1, rec->block_num,
+		     &rec->rec_det);
+
+out:
+	kfree(rec);
+}
+
+
+void mlx4_ib_init_alias_guid_work(struct mlx4_ib_dev *dev, int port)
+{
+	unsigned long flags, flags1;
+
+	if (!mlx4_is_master(dev->dev))
+		return;
+	spin_lock_irqsave(&dev->sriov.going_down_lock, flags);
+	spin_lock_irqsave(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	if (!dev->sriov.is_going_down) {
+		queue_delayed_work(dev->sriov.alias_guid.ports_guid[port].wq,
+			   &dev->sriov.alias_guid.ports_guid[port].alias_guid_work, 0);
+	}
+	spin_unlock_irqrestore(&dev->sriov.alias_guid.ag_work_lock, flags1);
+	spin_unlock_irqrestore(&dev->sriov.going_down_lock, flags);
+}
+
+void mlx4_ib_destroy_alias_guid_service(struct mlx4_ib_dev *dev)
+{
+	int i;
+	struct mlx4_ib_sriov *sriov = &dev->sriov;
+	struct mlx4_alias_guid_work_context *cb_ctx;
+	struct mlx4_sriov_alias_guid_port_rec_det *det;
+	struct ib_sa_query *sa_query;
+	unsigned long flags;
+
+	for (i = 0 ; i < dev->num_ports; i++) {
+		cancel_delayed_work(&dev->sriov.alias_guid.ports_guid[i].alias_guid_work);
+		det = &sriov->alias_guid.ports_guid[i];
+		spin_lock_irqsave(&sriov->alias_guid.ag_work_lock, flags);
+		while (!list_empty(&det->cb_list)) {
+			cb_ctx = list_entry(det->cb_list.next,
+					    struct mlx4_alias_guid_work_context,
+					    list);
+			sa_query = cb_ctx->sa_query;
+			cb_ctx->sa_query = NULL;
+			list_del(&cb_ctx->list);
+			spin_unlock_irqrestore(&sriov->alias_guid.ag_work_lock, flags);
+			ib_sa_cancel_query(cb_ctx->query_id, sa_query);
+			wait_for_completion(&cb_ctx->done);
+			kfree(cb_ctx);
+			spin_lock_irqsave(&sriov->alias_guid.ag_work_lock, flags);
+		}
+		spin_unlock_irqrestore(&sriov->alias_guid.ag_work_lock, flags);
+	}
+	for (i = 0 ; i < dev->num_ports; i++) {
+		flush_workqueue(dev->sriov.alias_guid.ports_guid[i].wq);
+		destroy_workqueue(dev->sriov.alias_guid.ports_guid[i].wq);
+	}
+	ib_sa_unregister_client(dev->sriov.alias_guid.sa_client);
+	kfree(dev->sriov.alias_guid.sa_client);
+}
+
+int mlx4_ib_init_alias_guid_service(struct mlx4_ib_dev *dev)
+{
+	char alias_wq_name[15];
+	int ret = 0;
+	int i, j, k;
+	union ib_gid gid;
+
+	if (!mlx4_is_master(dev->dev))
+		return 0;
+	dev->sriov.alias_guid.sa_client =
+		kzalloc(sizeof *dev->sriov.alias_guid.sa_client, GFP_KERNEL);
+	if (!dev->sriov.alias_guid.sa_client)
+		return -ENOMEM;
+
+	ib_sa_register_client(dev->sriov.alias_guid.sa_client);
+
+	spin_lock_init(&dev->sriov.alias_guid.ag_work_lock);
+
+	for (i = 1; i <= dev->num_ports; ++i) {
+		if (dev->ib_dev.query_gid(&dev->ib_dev , i, 0, &gid)) {
+			ret = -EFAULT;
+			goto err_unregister;
+		}
+	}
+
+	for (i = 0 ; i < dev->num_ports; i++) {
+		memset(&dev->sriov.alias_guid.ports_guid[i], 0,
+		       sizeof (struct mlx4_sriov_alias_guid_port_rec_det));
+		/*Check if the SM doesn't need to assign the GUIDs*/
+		for (j = 0; j < NUM_ALIAS_GUID_REC_IN_PORT; j++) {
+			if (mlx4_ib_sm_guid_assign) {
+				dev->sriov.alias_guid.ports_guid[i].
+					all_rec_per_port[j].
+					ownership = MLX4_GUID_DRIVER_ASSIGN;
+				continue;
+			}
+			dev->sriov.alias_guid.ports_guid[i].all_rec_per_port[j].
+					ownership = MLX4_GUID_NONE_ASSIGN;
+			/*mark each val as it was deleted,
+			  till the sysAdmin will give it valid val*/
+			for (k = 0; k < NUM_ALIAS_GUID_IN_REC; k++) {
+				*(__be64 *)&dev->sriov.alias_guid.ports_guid[i].
+					all_rec_per_port[j].all_recs[GUID_REC_SIZE * k] =
+						cpu_to_be64(MLX4_GUID_FOR_DELETE_VAL);
+			}
+		}
+		INIT_LIST_HEAD(&dev->sriov.alias_guid.ports_guid[i].cb_list);
+		/*prepare the records, set them to be allocated by sm*/
+		for (j = 0 ; j < NUM_ALIAS_GUID_REC_IN_PORT; j++)
+			invalidate_guid_record(dev, i + 1, j);
+
+		dev->sriov.alias_guid.ports_guid[i].parent = &dev->sriov.alias_guid;
+		dev->sriov.alias_guid.ports_guid[i].port  = i;
+		if (mlx4_ib_sm_guid_assign)
+			set_all_slaves_guids(dev, i);
+
+		snprintf(alias_wq_name, sizeof alias_wq_name, "alias_guid%d", i);
+		dev->sriov.alias_guid.ports_guid[i].wq =
+			create_singlethread_workqueue(alias_wq_name);
+		if (!dev->sriov.alias_guid.ports_guid[i].wq) {
+			ret = -ENOMEM;
+			goto err_thread;
+		}
+		INIT_DELAYED_WORK(&dev->sriov.alias_guid.ports_guid[i].alias_guid_work,
+			  alias_guid_work);
+	}
+	return 0;
+
+err_thread:
+	for (--i; i >= 0; i--) {
+		destroy_workqueue(dev->sriov.alias_guid.ports_guid[i].wq);
+		dev->sriov.alias_guid.ports_guid[i].wq = NULL;
+	}
+
+err_unregister:
+	ib_sa_unregister_client(dev->sriov.alias_guid.sa_client);
+	kfree(dev->sriov.alias_guid.sa_client);
+	dev->sriov.alias_guid.sa_client = NULL;
+	pr_err("init_alias_guid_service: Failed. (ret:%d)\n", ret);
+	return ret;
+}

drivers/infiniband/hw/mlx4/cm.c

@@ -0,0 +1,432 @@
+/*
+ * Copyright (c) 2012 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 <rdma/ib_mad.h>
+
+#include <linux/mlx4/cmd.h>
+#include <linux/rbtree.h>
+#include <linux/idr.h>
+#include <rdma/ib_cm.h>
+
+#include "mlx4_ib.h"
+
+#define CM_CLEANUP_CACHE_TIMEOUT  (5 * HZ)
+
+struct id_map_entry {
+	struct rb_node node;
+
+	u32 sl_cm_id;
+	u32 pv_cm_id;
+	int slave_id;
+	int scheduled_delete;
+	struct mlx4_ib_dev *dev;
+
+	struct list_head list;
+	struct delayed_work timeout;
+};
+
+struct cm_generic_msg {
+	struct ib_mad_hdr hdr;
+
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+};
+
+struct cm_req_msg {
+	unsigned char unused[0x60];
+	union ib_gid primary_path_sgid;
+};
+
+
+static void set_local_comm_id(struct ib_mad *mad, u32 cm_id)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+	msg->local_comm_id = cpu_to_be32(cm_id);
+}
+
+static u32 get_local_comm_id(struct ib_mad *mad)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+
+	return be32_to_cpu(msg->local_comm_id);
+}
+
+static void set_remote_comm_id(struct ib_mad *mad, u32 cm_id)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+	msg->remote_comm_id = cpu_to_be32(cm_id);
+}
+
+static u32 get_remote_comm_id(struct ib_mad *mad)
+{
+	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+
+	return be32_to_cpu(msg->remote_comm_id);
+}
+
+static union ib_gid gid_from_req_msg(struct ib_device *ibdev, struct ib_mad *mad)
+{
+	struct cm_req_msg *msg = (struct cm_req_msg *)mad;
+
+	return msg->primary_path_sgid;
+}
+
+/* Lock should be taken before called */
+static struct id_map_entry *
+id_map_find_by_sl_id(struct ib_device *ibdev, u32 slave_id, u32 sl_cm_id)
+{
+	struct rb_root *sl_id_map = &to_mdev(ibdev)->sriov.sl_id_map;
+	struct rb_node *node = sl_id_map->rb_node;
+
+	while (node) {
+		struct id_map_entry *id_map_entry =
+			rb_entry(node, struct id_map_entry, node);
+
+		if (id_map_entry->sl_cm_id > sl_cm_id)
+			node = node->rb_left;
+		else if (id_map_entry->sl_cm_id < sl_cm_id)
+			node = node->rb_right;
+		else if (id_map_entry->slave_id > slave_id)
+			node = node->rb_left;
+		else if (id_map_entry->slave_id < slave_id)
+			node = node->rb_right;
+		else
+			return id_map_entry;
+	}
+	return NULL;
+}
+
+static void id_map_ent_timeout(struct work_struct *work)
+{
+	struct delayed_work *delay = to_delayed_work(work);
+	struct id_map_entry *ent = container_of(delay, struct id_map_entry, timeout);
+	struct id_map_entry *db_ent, *found_ent;
+	struct mlx4_ib_dev *dev = ent->dev;
+	struct mlx4_ib_sriov *sriov = &dev->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	int pv_id = (int) ent->pv_cm_id;
+
+	spin_lock(&sriov->id_map_lock);
+	db_ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, pv_id);
+	if (!db_ent)
+		goto out;
+	found_ent = id_map_find_by_sl_id(&dev->ib_dev, ent->slave_id, ent->sl_cm_id);
+	if (found_ent && found_ent == ent)
+		rb_erase(&found_ent->node, sl_id_map);
+	idr_remove(&sriov->pv_id_table, pv_id);
+
+out:
+	list_del(&ent->list);
+	spin_unlock(&sriov->id_map_lock);
+	kfree(ent);
+}
+
+static void id_map_find_del(struct ib_device *ibdev, int pv_cm_id)
+{
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	struct id_map_entry *ent, *found_ent;
+
+	spin_lock(&sriov->id_map_lock);
+	ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, pv_cm_id);
+	if (!ent)
+		goto out;
+	found_ent = id_map_find_by_sl_id(ibdev, ent->slave_id, ent->sl_cm_id);
+	if (found_ent && found_ent == ent)
+		rb_erase(&found_ent->node, sl_id_map);
+	idr_remove(&sriov->pv_id_table, pv_cm_id);
+out:
+	spin_unlock(&sriov->id_map_lock);
+}
+
+static void sl_id_map_add(struct ib_device *ibdev, struct id_map_entry *new)
+{
+	struct rb_root *sl_id_map = &to_mdev(ibdev)->sriov.sl_id_map;
+	struct rb_node **link = &sl_id_map->rb_node, *parent = NULL;
+	struct id_map_entry *ent;
+	int slave_id = new->slave_id;
+	int sl_cm_id = new->sl_cm_id;
+
+	ent = id_map_find_by_sl_id(ibdev, slave_id, sl_cm_id);
+	if (ent) {
+		pr_debug("overriding existing sl_id_map entry (cm_id = %x)\n",
+			 sl_cm_id);
+
+		rb_replace_node(&ent->node, &new->node, sl_id_map);
+		return;
+	}
+
+	/* Go to the bottom of the tree */
+	while (*link) {
+		parent = *link;
+		ent = rb_entry(parent, struct id_map_entry, node);
+
+		if (ent->sl_cm_id > sl_cm_id || (ent->sl_cm_id == sl_cm_id && ent->slave_id > slave_id))
+			link = &(*link)->rb_left;
+		else
+			link = &(*link)->rb_right;
+	}
+
+	rb_link_node(&new->node, parent, link);
+	rb_insert_color(&new->node, sl_id_map);
+}
+
+static struct id_map_entry *
+id_map_alloc(struct ib_device *ibdev, int slave_id, u32 sl_cm_id)
+{
+	int ret;
+	struct id_map_entry *ent;
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+
+	ent = kmalloc(sizeof (struct id_map_entry), GFP_KERNEL);
+	if (!ent) {
+		mlx4_ib_warn(ibdev, "Couldn't allocate id cache entry - out of memory\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	ent->sl_cm_id = sl_cm_id;
+	ent->slave_id = slave_id;
+	ent->scheduled_delete = 0;
+	ent->dev = to_mdev(ibdev);
+	INIT_DELAYED_WORK(&ent->timeout, id_map_ent_timeout);
+
+	idr_preload(GFP_KERNEL);
+	spin_lock(&to_mdev(ibdev)->sriov.id_map_lock);
+
+	ret = idr_alloc_cyclic(&sriov->pv_id_table, ent, 0, 0, GFP_NOWAIT);
+	if (ret >= 0) {
+		ent->pv_cm_id = (u32)ret;
+		sl_id_map_add(ibdev, ent);
+		list_add_tail(&ent->list, &sriov->cm_list);
+	}
+
+	spin_unlock(&sriov->id_map_lock);
+	idr_preload_end();
+
+	if (ret >= 0)
+		return ent;
+
+	/*error flow*/
+	kfree(ent);
+	mlx4_ib_warn(ibdev, "No more space in the idr (err:0x%x)\n", ret);
+	return ERR_PTR(-ENOMEM);
+}
+
+static struct id_map_entry *
+id_map_get(struct ib_device *ibdev, int *pv_cm_id, int sl_cm_id, int slave_id)
+{
+	struct id_map_entry *ent;
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+
+	spin_lock(&sriov->id_map_lock);
+	if (*pv_cm_id == -1) {
+		ent = id_map_find_by_sl_id(ibdev, sl_cm_id, slave_id);
+		if (ent)
+			*pv_cm_id = (int) ent->pv_cm_id;
+	} else
+		ent = (struct id_map_entry *)idr_find(&sriov->pv_id_table, *pv_cm_id);
+	spin_unlock(&sriov->id_map_lock);
+
+	return ent;
+}
+
+static void schedule_delayed(struct ib_device *ibdev, struct id_map_entry *id)
+{
+	struct mlx4_ib_sriov *sriov = &to_mdev(ibdev)->sriov;
+	unsigned long flags;
+
+	spin_lock(&sriov->id_map_lock);
+	spin_lock_irqsave(&sriov->going_down_lock, flags);
+	/*make sure that there is no schedule inside the scheduled work.*/
+	if (!sriov->is_going_down) {
+		id->scheduled_delete = 1;
+		schedule_delayed_work(&id->timeout, CM_CLEANUP_CACHE_TIMEOUT);
+	}
+	spin_unlock_irqrestore(&sriov->going_down_lock, flags);
+	spin_unlock(&sriov->id_map_lock);
+}
+
+int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id,
+		struct ib_mad *mad)
+{
+	struct id_map_entry *id;
+	u32 sl_cm_id;
+	int pv_cm_id = -1;
+
+	sl_cm_id = get_local_comm_id(mad);
+
+	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID ||
+			mad->mad_hdr.attr_id == CM_REP_ATTR_ID) {
+		id = id_map_alloc(ibdev, slave_id, sl_cm_id);
+		if (IS_ERR(id)) {
+			mlx4_ib_warn(ibdev, "%s: id{slave: %d, sl_cm_id: 0x%x} Failed to id_map_alloc\n",
+				__func__, slave_id, sl_cm_id);
+			return PTR_ERR(id);
+		}
+	} else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID) {
+		return 0;
+	} else {
+		id = id_map_get(ibdev, &pv_cm_id, slave_id, sl_cm_id);
+	}
+
+	if (!id) {
+		pr_debug("id{slave: %d, sl_cm_id: 0x%x} is NULL!\n",
+			 slave_id, sl_cm_id);
+		return -EINVAL;
+	}
+
+	set_local_comm_id(mad, id->pv_cm_id);
+
+	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
+		schedule_delayed(ibdev, id);
+	else if (mad->mad_hdr.attr_id == CM_DREP_ATTR_ID)
+		id_map_find_del(ibdev, pv_cm_id);
+
+	return 0;
+}
+
+int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
+							     struct ib_mad *mad)
+{
+	u32 pv_cm_id;
+	struct id_map_entry *id;
+
+	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID) {
+		union ib_gid gid;
+
+		gid = gid_from_req_msg(ibdev, mad);
+		*slave = mlx4_ib_find_real_gid(ibdev, port, gid.global.interface_id);
+		if (*slave < 0) {
+			mlx4_ib_warn(ibdev, "failed matching slave_id by gid (0x%llx)\n",
+					gid.global.interface_id);
+			return -ENOENT;
+		}
+		return 0;
+	}
+
+	pv_cm_id = get_remote_comm_id(mad);
+	id = id_map_get(ibdev, (int *)&pv_cm_id, -1, -1);
+
+	if (!id) {
+		pr_debug("Couldn't find an entry for pv_cm_id 0x%x\n", pv_cm_id);
+		return -ENOENT;
+	}
+
+	*slave = id->slave_id;
+	set_remote_comm_id(mad, id->sl_cm_id);
+
+	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
+		schedule_delayed(ibdev, id);
+	else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID ||
+			mad->mad_hdr.attr_id == CM_DREP_ATTR_ID) {
+		id_map_find_del(ibdev, (int) pv_cm_id);
+	}
+
+	return 0;
+}
+
+void mlx4_ib_cm_paravirt_init(struct mlx4_ib_dev *dev)
+{
+	spin_lock_init(&dev->sriov.id_map_lock);
+	INIT_LIST_HEAD(&dev->sriov.cm_list);
+	dev->sriov.sl_id_map = RB_ROOT;
+	idr_init(&dev->sriov.pv_id_table);
+}
+
+/* slave = -1 ==> all slaves */
+/* TBD -- call paravirt clean for single slave.  Need for slave RESET event */
+void mlx4_ib_cm_paravirt_clean(struct mlx4_ib_dev *dev, int slave)
+{
+	struct mlx4_ib_sriov *sriov = &dev->sriov;
+	struct rb_root *sl_id_map = &sriov->sl_id_map;
+	struct list_head lh;
+	struct rb_node *nd;
+	int need_flush = 1;
+	struct id_map_entry *map, *tmp_map;
+	/* cancel all delayed work queue entries */
+	INIT_LIST_HEAD(&lh);
+	spin_lock(&sriov->id_map_lock);
+	list_for_each_entry_safe(map, tmp_map, &dev->sriov.cm_list, list) {
+		if (slave < 0 || slave == map->slave_id) {
+			if (map->scheduled_delete)
+				need_flush &= !!cancel_delayed_work(&map->timeout);
+		}
+	}
+
+	spin_unlock(&sriov->id_map_lock);
+
+	if (!need_flush)
+		flush_scheduled_work(); /* make sure all timers were flushed */
+
+	/* now, remove all leftover entries from databases*/
+	spin_lock(&sriov->id_map_lock);
+	if (slave < 0) {
+		while (rb_first(sl_id_map)) {
+			struct id_map_entry *ent =
+				rb_entry(rb_first(sl_id_map),
+					 struct id_map_entry, node);
+
+			rb_erase(&ent->node, sl_id_map);
+			idr_remove(&sriov->pv_id_table, (int) ent->pv_cm_id);
+		}
+		list_splice_init(&dev->sriov.cm_list, &lh);
+	} else {
+		/* first, move nodes belonging to slave to db remove list */
+		nd = rb_first(sl_id_map);
+		while (nd) {
+			struct id_map_entry *ent =
+				rb_entry(nd, struct id_map_entry, node);
+			nd = rb_next(nd);
+			if (ent->slave_id == slave)
+				list_move_tail(&ent->list, &lh);
+		}
+		/* remove those nodes from databases */
+		list_for_each_entry_safe(map, tmp_map, &lh, list) {
+			rb_erase(&map->node, sl_id_map);
+			idr_remove(&sriov->pv_id_table, (int) map->pv_cm_id);
+		}
+
+		/* add remaining nodes from cm_list */
+		list_for_each_entry_safe(map, tmp_map, &dev->sriov.cm_list, list) {
+			if (slave == map->slave_id)
+				list_move_tail(&map->list, &lh);
+		}
+	}
+
+	spin_unlock(&sriov->id_map_lock);
+
+	/* free any map entries left behind due to cancel_delayed_work above */
+	list_for_each_entry_safe(map, tmp_map, &lh, list) {
+		list_del(&map->list);
+		kfree(map);
+	}
+}

drivers/infiniband/hw/mlx4/cq.c

@@ -0,0 +1,896 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 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/mlx4/cq.h>
+#include <linux/mlx4/qp.h>
+#include <linux/mlx4/srq.h>
+#include <linux/slab.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+static void mlx4_ib_cq_comp(struct mlx4_cq *cq)
+{
+	struct ib_cq *ibcq = &to_mibcq(cq)->ibcq;
+	ibcq->comp_handler(ibcq, ibcq->cq_context);
+}
+
+static void mlx4_ib_cq_event(struct mlx4_cq *cq, enum mlx4_event type)
+{
+	struct ib_event event;
+	struct ib_cq *ibcq;
+
+	if (type != MLX4_EVENT_TYPE_CQ_ERROR) {
+		pr_warn("Unexpected event type %d "
+		       "on CQ %06x\n", type, cq->cqn);
+		return;
+	}
+
+	ibcq = &to_mibcq(cq)->ibcq;
+	if (ibcq->event_handler) {
+		event.device     = ibcq->device;
+		event.event      = IB_EVENT_CQ_ERR;
+		event.element.cq = ibcq;
+		ibcq->event_handler(&event, ibcq->cq_context);
+	}
+}
+
+static void *get_cqe_from_buf(struct mlx4_ib_cq_buf *buf, int n)
+{
+	return mlx4_buf_offset(&buf->buf, n * buf->entry_size);
+}
+
+static void *get_cqe(struct mlx4_ib_cq *cq, int n)
+{
+	return get_cqe_from_buf(&cq->buf, n);
+}
+
+static void *get_sw_cqe(struct mlx4_ib_cq *cq, int n)
+{
+	struct mlx4_cqe *cqe = get_cqe(cq, n & cq->ibcq.cqe);
+	struct mlx4_cqe *tcqe = ((cq->buf.entry_size == 64) ? (cqe + 1) : cqe);
+
+	return (!!(tcqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK) ^
+		!!(n & (cq->ibcq.cqe + 1))) ? NULL : cqe;
+}
+
+static struct mlx4_cqe *next_cqe_sw(struct mlx4_ib_cq *cq)
+{
+	return get_sw_cqe(cq, cq->mcq.cons_index);
+}
+
+int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
+{
+	struct mlx4_ib_cq *mcq = to_mcq(cq);
+	struct mlx4_ib_dev *dev = to_mdev(cq->device);
+
+	return mlx4_cq_modify(dev->dev, &mcq->mcq, cq_count, cq_period);
+}
+
+static int mlx4_ib_alloc_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int nent)
+{
+	int err;
+
+	err = mlx4_buf_alloc(dev->dev, nent * dev->dev->caps.cqe_size,
+			     PAGE_SIZE * 2, &buf->buf);
+
+	if (err)
+		goto out;
+
+	buf->entry_size = dev->dev->caps.cqe_size;
+	err = mlx4_mtt_init(dev->dev, buf->buf.npages, buf->buf.page_shift,
+				    &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_buf_write_mtt(dev->dev, &buf->mtt, &buf->buf);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	mlx4_buf_free(dev->dev, nent * buf->entry_size, &buf->buf);
+
+out:
+	return err;
+}
+
+static void mlx4_ib_free_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int cqe)
+{
+	mlx4_buf_free(dev->dev, (cqe + 1) * buf->entry_size, &buf->buf);
+}
+
+static int mlx4_ib_get_cq_umem(struct mlx4_ib_dev *dev, struct ib_ucontext *context,
+			       struct mlx4_ib_cq_buf *buf, struct ib_umem **umem,
+			       u64 buf_addr, int cqe)
+{
+	int err;
+	int cqe_size = dev->dev->caps.cqe_size;
+
+	*umem = ib_umem_get(context, buf_addr, cqe * cqe_size,
+			    IB_ACCESS_LOCAL_WRITE, 1);
+	if (IS_ERR(*umem))
+		return PTR_ERR(*umem);
+
+	err = mlx4_mtt_init(dev->dev, ib_umem_page_count(*umem),
+			    ilog2((*umem)->page_size), &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_ib_umem_write_mtt(dev, &buf->mtt, *umem);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	ib_umem_release(*umem);
+
+	return err;
+}
+
+struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
+				struct ib_ucontext *context,
+				struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct mlx4_ib_cq *cq;
+	struct mlx4_uar *uar;
+	int err;
+
+	if (entries < 1 || entries > dev->dev->caps.max_cqes)
+		return ERR_PTR(-EINVAL);
+
+	cq = kmalloc(sizeof *cq, GFP_KERNEL);
+	if (!cq)
+		return ERR_PTR(-ENOMEM);
+
+	entries      = roundup_pow_of_two(entries + 1);
+	cq->ibcq.cqe = entries - 1;
+	mutex_init(&cq->resize_mutex);
+	spin_lock_init(&cq->lock);
+	cq->resize_buf = NULL;
+	cq->resize_umem = NULL;
+
+	if (context) {
+		struct mlx4_ib_create_cq ucmd;
+
+		if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
+			err = -EFAULT;
+			goto err_cq;
+		}
+
+		err = mlx4_ib_get_cq_umem(dev, context, &cq->buf, &cq->umem,
+					  ucmd.buf_addr, entries);
+		if (err)
+			goto err_cq;
+
+		err = mlx4_ib_db_map_user(to_mucontext(context), ucmd.db_addr,
+					  &cq->db);
+		if (err)
+			goto err_mtt;
+
+		uar = &to_mucontext(context)->uar;
+	} else {
+		err = mlx4_db_alloc(dev->dev, &cq->db, 1);
+		if (err)
+			goto err_cq;
+
+		cq->mcq.set_ci_db  = cq->db.db;
+		cq->mcq.arm_db     = cq->db.db + 1;
+		*cq->mcq.set_ci_db = 0;
+		*cq->mcq.arm_db    = 0;
+
+		err = mlx4_ib_alloc_cq_buf(dev, &cq->buf, entries);
+		if (err)
+			goto err_db;
+
+		uar = &dev->priv_uar;
+	}
+
+	if (dev->eq_table)
+		vector = dev->eq_table[vector % ibdev->num_comp_vectors];
+
+	err = mlx4_cq_alloc(dev->dev, entries, &cq->buf.mtt, uar,
+			    cq->db.dma, &cq->mcq, vector, 0, 0);
+	if (err)
+		goto err_dbmap;
+
+	cq->mcq.comp  = mlx4_ib_cq_comp;
+	cq->mcq.event = mlx4_ib_cq_event;
+
+	if (context)
+		if (ib_copy_to_udata(udata, &cq->mcq.cqn, sizeof (__u32))) {
+			err = -EFAULT;
+			goto err_dbmap;
+		}
+
+	return &cq->ibcq;
+
+err_dbmap:
+	if (context)
+		mlx4_ib_db_unmap_user(to_mucontext(context), &cq->db);
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
+
+	if (context)
+		ib_umem_release(cq->umem);
+	else
+		mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
+
+err_db:
+	if (!context)
+		mlx4_db_free(dev->dev, &cq->db);
+
+err_cq:
+	kfree(cq);
+
+	return ERR_PTR(err);
+}
+
+static int mlx4_alloc_resize_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				  int entries)
+{
+	int err;
+
+	if (cq->resize_buf)
+		return -EBUSY;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_alloc_cq_buf(dev, &cq->resize_buf->buf, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_alloc_resize_umem(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				   int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_resize_cq ucmd;
+	int err;
+
+	if (cq->resize_umem)
+		return -EBUSY;
+
+	if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd))
+		return -EFAULT;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_get_cq_umem(dev, cq->umem->context, &cq->resize_buf->buf,
+				  &cq->resize_umem, ucmd.buf_addr, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_ib_get_outstanding_cqes(struct mlx4_ib_cq *cq)
+{
+	u32 i;
+
+	i = cq->mcq.cons_index;
+	while (get_sw_cqe(cq, i & cq->ibcq.cqe))
+		++i;
+
+	return i - cq->mcq.cons_index;
+}
+
+static void mlx4_ib_cq_resize_copy_cqes(struct mlx4_ib_cq *cq)
+{
+	struct mlx4_cqe *cqe, *new_cqe;
+	int i;
+	int cqe_size = cq->buf.entry_size;
+	int cqe_inc = cqe_size == 64 ? 1 : 0;
+
+	i = cq->mcq.cons_index;
+	cqe = get_cqe(cq, i & cq->ibcq.cqe);
+	cqe += cqe_inc;
+
+	while ((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) != MLX4_CQE_OPCODE_RESIZE) {
+		new_cqe = get_cqe_from_buf(&cq->resize_buf->buf,
+					   (i + 1) & cq->resize_buf->cqe);
+		memcpy(new_cqe, get_cqe(cq, i & cq->ibcq.cqe), cqe_size);
+		new_cqe += cqe_inc;
+
+		new_cqe->owner_sr_opcode = (cqe->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK) |
+			(((i + 1) & (cq->resize_buf->cqe + 1)) ? MLX4_CQE_OWNER_MASK : 0);
+		cqe = get_cqe(cq, ++i & cq->ibcq.cqe);
+		cqe += cqe_inc;
+	}
+	++cq->mcq.cons_index;
+}
+
+int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibcq->device);
+	struct mlx4_ib_cq *cq = to_mcq(ibcq);
+	struct mlx4_mtt mtt;
+	int outst_cqe;
+	int err;
+
+	mutex_lock(&cq->resize_mutex);
+
+	if (entries < 1 || entries > dev->dev->caps.max_cqes) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	entries = roundup_pow_of_two(entries + 1);
+	if (entries == ibcq->cqe + 1) {
+		err = 0;
+		goto out;
+	}
+
+	if (ibcq->uobject) {
+		err = mlx4_alloc_resize_umem(dev, cq, entries, udata);
+		if (err)
+			goto out;
+	} else {
+		/* Can't be smaller than the number of outstanding CQEs */
+		outst_cqe = mlx4_ib_get_outstanding_cqes(cq);
+		if (entries < outst_cqe + 1) {
+			err = 0;
+			goto out;
+		}
+
+		err = mlx4_alloc_resize_buf(dev, cq, entries);
+		if (err)
+			goto out;
+	}
+
+	mtt = cq->buf.mtt;
+
+	err = mlx4_cq_resize(dev->dev, &cq->mcq, entries, &cq->resize_buf->buf.mtt);
+	if (err)
+		goto err_buf;
+
+	mlx4_mtt_cleanup(dev->dev, &mtt);
+	if (ibcq->uobject) {
+		cq->buf      = cq->resize_buf->buf;
+		cq->ibcq.cqe = cq->resize_buf->cqe;
+		ib_umem_release(cq->umem);
+		cq->umem     = cq->resize_umem;
+
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		cq->resize_umem = NULL;
+	} else {
+		struct mlx4_ib_cq_buf tmp_buf;
+		int tmp_cqe = 0;
+
+		spin_lock_irq(&cq->lock);
+		if (cq->resize_buf) {
+			mlx4_ib_cq_resize_copy_cqes(cq);
+			tmp_buf = cq->buf;
+			tmp_cqe = cq->ibcq.cqe;
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+		spin_unlock_irq(&cq->lock);
+
+		if (tmp_cqe)
+			mlx4_ib_free_cq_buf(dev, &tmp_buf, tmp_cqe);
+	}
+
+	goto out;
+
+err_buf:
+	mlx4_mtt_cleanup(dev->dev, &cq->resize_buf->buf.mtt);
+	if (!ibcq->uobject)
+		mlx4_ib_free_cq_buf(dev, &cq->resize_buf->buf,
+				    cq->resize_buf->cqe);
+
+	kfree(cq->resize_buf);
+	cq->resize_buf = NULL;
+
+	if (cq->resize_umem) {
+		ib_umem_release(cq->resize_umem);
+		cq->resize_umem = NULL;
+	}
+
+out:
+	mutex_unlock(&cq->resize_mutex);
+
+	return err;
+}
+
+int mlx4_ib_destroy_cq(struct ib_cq *cq)
+{
+	struct mlx4_ib_dev *dev = to_mdev(cq->device);
+	struct mlx4_ib_cq *mcq = to_mcq(cq);
+
+	mlx4_cq_free(dev->dev, &mcq->mcq);
+	mlx4_mtt_cleanup(dev->dev, &mcq->buf.mtt);
+
+	if (cq->uobject) {
+		mlx4_ib_db_unmap_user(to_mucontext(cq->uobject->context), &mcq->db);
+		ib_umem_release(mcq->umem);
+	} else {
+		mlx4_ib_free_cq_buf(dev, &mcq->buf, cq->cqe);
+		mlx4_db_free(dev->dev, &mcq->db);
+	}
+
+	kfree(mcq);
+
+	return 0;
+}
+
+static void dump_cqe(void *cqe)
+{
+	__be32 *buf = cqe;
+
+	pr_debug("CQE contents %08x %08x %08x %08x %08x %08x %08x %08x\n",
+	       be32_to_cpu(buf[0]), be32_to_cpu(buf[1]), be32_to_cpu(buf[2]),
+	       be32_to_cpu(buf[3]), be32_to_cpu(buf[4]), be32_to_cpu(buf[5]),
+	       be32_to_cpu(buf[6]), be32_to_cpu(buf[7]));
+}
+
+static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
+				     struct ib_wc *wc)
+{
+	if (cqe->syndrome == MLX4_CQE_SYNDROME_LOCAL_QP_OP_ERR) {
+		pr_debug("local QP operation err "
+		       "(QPN %06x, WQE index %x, vendor syndrome %02x, "
+		       "opcode = %02x)\n",
+		       be32_to_cpu(cqe->my_qpn), be16_to_cpu(cqe->wqe_index),
+		       cqe->vendor_err_syndrome,
+		       cqe->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK);
+		dump_cqe(cqe);
+	}
+
+	switch (cqe->syndrome) {
+	case MLX4_CQE_SYNDROME_LOCAL_LENGTH_ERR:
+		wc->status = IB_WC_LOC_LEN_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_QP_OP_ERR:
+		wc->status = IB_WC_LOC_QP_OP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_PROT_ERR:
+		wc->status = IB_WC_LOC_PROT_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_WR_FLUSH_ERR:
+		wc->status = IB_WC_WR_FLUSH_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_MW_BIND_ERR:
+		wc->status = IB_WC_MW_BIND_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_BAD_RESP_ERR:
+		wc->status = IB_WC_BAD_RESP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_LOCAL_ACCESS_ERR:
+		wc->status = IB_WC_LOC_ACCESS_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_INVAL_REQ_ERR:
+		wc->status = IB_WC_REM_INV_REQ_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_ACCESS_ERR:
+		wc->status = IB_WC_REM_ACCESS_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_OP_ERR:
+		wc->status = IB_WC_REM_OP_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_TRANSPORT_RETRY_EXC_ERR:
+		wc->status = IB_WC_RETRY_EXC_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_RNR_RETRY_EXC_ERR:
+		wc->status = IB_WC_RNR_RETRY_EXC_ERR;
+		break;
+	case MLX4_CQE_SYNDROME_REMOTE_ABORTED_ERR:
+		wc->status = IB_WC_REM_ABORT_ERR;
+		break;
+	default:
+		wc->status = IB_WC_GENERAL_ERR;
+		break;
+	}
+
+	wc->vendor_err = cqe->vendor_err_syndrome;
+}
+
+static int mlx4_ib_ipoib_csum_ok(__be16 status, __be16 checksum)
+{
+	return ((status & cpu_to_be16(MLX4_CQE_STATUS_IPV4      |
+				      MLX4_CQE_STATUS_IPV4F     |
+				      MLX4_CQE_STATUS_IPV4OPT   |
+				      MLX4_CQE_STATUS_IPV6      |
+				      MLX4_CQE_STATUS_IPOK)) ==
+		cpu_to_be16(MLX4_CQE_STATUS_IPV4        |
+			    MLX4_CQE_STATUS_IPOK))              &&
+		(status & cpu_to_be16(MLX4_CQE_STATUS_UDP       |
+				      MLX4_CQE_STATUS_TCP))     &&
+		checksum == cpu_to_be16(0xffff);
+}
+
+static int use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct ib_wc *wc,
+			   unsigned tail, struct mlx4_cqe *cqe)
+{
+	struct mlx4_ib_proxy_sqp_hdr *hdr;
+
+	ib_dma_sync_single_for_cpu(qp->ibqp.device,
+				   qp->sqp_proxy_rcv[tail].map,
+				   sizeof (struct mlx4_ib_proxy_sqp_hdr),
+				   DMA_FROM_DEVICE);
+	hdr = (struct mlx4_ib_proxy_sqp_hdr *) (qp->sqp_proxy_rcv[tail].addr);
+	wc->pkey_index	= be16_to_cpu(hdr->tun.pkey_index);
+	wc->slid	= be16_to_cpu(hdr->tun.slid_mac_47_32);
+	wc->sl		= (u8) (be16_to_cpu(hdr->tun.sl_vid) >> 12);
+	wc->src_qp	= be32_to_cpu(hdr->tun.flags_src_qp) & 0xFFFFFF;
+	wc->wc_flags   |= (hdr->tun.g_ml_path & 0x80) ? (IB_WC_GRH) : 0;
+	wc->dlid_path_bits = 0;
+
+	return 0;
+}
+
+static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
+			    struct mlx4_ib_qp **cur_qp,
+			    struct ib_wc *wc)
+{
+	struct mlx4_cqe *cqe;
+	struct mlx4_qp *mqp;
+	struct mlx4_ib_wq *wq;
+	struct mlx4_ib_srq *srq;
+	struct mlx4_srq *msrq = NULL;
+	int is_send;
+	int is_error;
+	u32 g_mlpath_rqpn;
+	u16 wqe_ctr;
+	unsigned tail = 0;
+
+repoll:
+	cqe = next_cqe_sw(cq);
+	if (!cqe)
+		return -EAGAIN;
+
+	if (cq->buf.entry_size == 64)
+		cqe++;
+
+	++cq->mcq.cons_index;
+
+	/*
+	 * Make sure we read CQ entry contents after we've checked the
+	 * ownership bit.
+	 */
+	rmb();
+
+	is_send  = cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK;
+	is_error = (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
+		MLX4_CQE_OPCODE_ERROR;
+
+	if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_OPCODE_NOP &&
+		     is_send)) {
+		pr_warn("Completion for NOP opcode detected!\n");
+		return -EINVAL;
+	}
+
+	/* Resize CQ in progress */
+	if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_RESIZE)) {
+		if (cq->resize_buf) {
+			struct mlx4_ib_dev *dev = to_mdev(cq->ibcq.device);
+
+			mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+
+		goto repoll;
+	}
+
+	if (!*cur_qp ||
+	    (be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK) != (*cur_qp)->mqp.qpn) {
+		/*
+		 * We do not have to take the QP table lock here,
+		 * because CQs will be locked while QPs are removed
+		 * from the table.
+		 */
+		mqp = __mlx4_qp_lookup(to_mdev(cq->ibcq.device)->dev,
+				       be32_to_cpu(cqe->vlan_my_qpn));
+		if (unlikely(!mqp)) {
+			pr_warn("CQ %06x with entry for unknown QPN %06x\n",
+			       cq->mcq.cqn, be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK);
+			return -EINVAL;
+		}
+
+		*cur_qp = to_mibqp(mqp);
+	}
+
+	wc->qp = &(*cur_qp)->ibqp;
+
+	if (wc->qp->qp_type == IB_QPT_XRC_TGT) {
+		u32 srq_num;
+		g_mlpath_rqpn = be32_to_cpu(cqe->g_mlpath_rqpn);
+		srq_num       = g_mlpath_rqpn & 0xffffff;
+		/* SRQ is also in the radix tree */
+		msrq = mlx4_srq_lookup(to_mdev(cq->ibcq.device)->dev,
+				       srq_num);
+		if (unlikely(!msrq)) {
+			pr_warn("CQ %06x with entry for unknown SRQN %06x\n",
+				cq->mcq.cqn, srq_num);
+			return -EINVAL;
+		}
+	}
+
+	if (is_send) {
+		wq = &(*cur_qp)->sq;
+		if (!(*cur_qp)->sq_signal_bits) {
+			wqe_ctr = be16_to_cpu(cqe->wqe_index);
+			wq->tail += (u16) (wqe_ctr - (u16) wq->tail);
+		}
+		wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)];
+		++wq->tail;
+	} else if ((*cur_qp)->ibqp.srq) {
+		srq = to_msrq((*cur_qp)->ibqp.srq);
+		wqe_ctr = be16_to_cpu(cqe->wqe_index);
+		wc->wr_id = srq->wrid[wqe_ctr];
+		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
+	} else if (msrq) {
+		srq = to_mibsrq(msrq);
+		wqe_ctr = be16_to_cpu(cqe->wqe_index);
+		wc->wr_id = srq->wrid[wqe_ctr];
+		mlx4_ib_free_srq_wqe(srq, wqe_ctr);
+	} else {
+		wq	  = &(*cur_qp)->rq;
+		tail	  = wq->tail & (wq->wqe_cnt - 1);
+		wc->wr_id = wq->wrid[tail];
+		++wq->tail;
+	}
+
+	if (unlikely(is_error)) {
+		mlx4_ib_handle_error_cqe((struct mlx4_err_cqe *) cqe, wc);
+		return 0;
+	}
+
+	wc->status = IB_WC_SUCCESS;
+
+	if (is_send) {
+		wc->wc_flags = 0;
+		switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
+		case MLX4_OPCODE_RDMA_WRITE_IMM:
+			wc->wc_flags |= IB_WC_WITH_IMM;
+		case MLX4_OPCODE_RDMA_WRITE:
+			wc->opcode    = IB_WC_RDMA_WRITE;
+			break;
+		case MLX4_OPCODE_SEND_IMM:
+			wc->wc_flags |= IB_WC_WITH_IMM;
+		case MLX4_OPCODE_SEND:
+		case MLX4_OPCODE_SEND_INVAL:
+			wc->opcode    = IB_WC_SEND;
+			break;
+		case MLX4_OPCODE_RDMA_READ:
+			wc->opcode    = IB_WC_RDMA_READ;
+			wc->byte_len  = be32_to_cpu(cqe->byte_cnt);
+			break;
+		case MLX4_OPCODE_ATOMIC_CS:
+			wc->opcode    = IB_WC_COMP_SWAP;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_ATOMIC_FA:
+			wc->opcode    = IB_WC_FETCH_ADD;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_MASKED_ATOMIC_CS:
+			wc->opcode    = IB_WC_MASKED_COMP_SWAP;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_MASKED_ATOMIC_FA:
+			wc->opcode    = IB_WC_MASKED_FETCH_ADD;
+			wc->byte_len  = 8;
+			break;
+		case MLX4_OPCODE_BIND_MW:
+			wc->opcode    = IB_WC_BIND_MW;
+			break;
+		case MLX4_OPCODE_LSO:
+			wc->opcode    = IB_WC_LSO;
+			break;
+		case MLX4_OPCODE_FMR:
+			wc->opcode    = IB_WC_FAST_REG_MR;
+			break;
+		case MLX4_OPCODE_LOCAL_INVAL:
+			wc->opcode    = IB_WC_LOCAL_INV;
+			break;
+		}
+	} else {
+		wc->byte_len = be32_to_cpu(cqe->byte_cnt);
+
+		switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
+		case MLX4_RECV_OPCODE_RDMA_WRITE_IMM:
+			wc->opcode	= IB_WC_RECV_RDMA_WITH_IMM;
+			wc->wc_flags	= IB_WC_WITH_IMM;
+			wc->ex.imm_data = cqe->immed_rss_invalid;
+			break;
+		case MLX4_RECV_OPCODE_SEND_INVAL:
+			wc->opcode	= IB_WC_RECV;
+			wc->wc_flags	= IB_WC_WITH_INVALIDATE;
+			wc->ex.invalidate_rkey = be32_to_cpu(cqe->immed_rss_invalid);
+			break;
+		case MLX4_RECV_OPCODE_SEND:
+			wc->opcode   = IB_WC_RECV;
+			wc->wc_flags = 0;
+			break;
+		case MLX4_RECV_OPCODE_SEND_IMM:
+			wc->opcode	= IB_WC_RECV;
+			wc->wc_flags	= IB_WC_WITH_IMM;
+			wc->ex.imm_data = cqe->immed_rss_invalid;
+			break;
+		}
+
+		if (mlx4_is_mfunc(to_mdev(cq->ibcq.device)->dev)) {
+			if ((*cur_qp)->mlx4_ib_qp_type &
+			    (MLX4_IB_QPT_PROXY_SMI_OWNER |
+			     MLX4_IB_QPT_PROXY_SMI | MLX4_IB_QPT_PROXY_GSI))
+				return use_tunnel_data(*cur_qp, cq, wc, tail, cqe);
+		}
+
+		wc->slid	   = be16_to_cpu(cqe->rlid);
+		g_mlpath_rqpn	   = be32_to_cpu(cqe->g_mlpath_rqpn);
+		wc->src_qp	   = g_mlpath_rqpn & 0xffffff;
+		wc->dlid_path_bits = (g_mlpath_rqpn >> 24) & 0x7f;
+		wc->wc_flags	  |= g_mlpath_rqpn & 0x80000000 ? IB_WC_GRH : 0;
+		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
+		wc->wc_flags	  |= mlx4_ib_ipoib_csum_ok(cqe->status,
+					cqe->checksum) ? IB_WC_IP_CSUM_OK : 0;
+		if (rdma_port_get_link_layer(wc->qp->device,
+				(*cur_qp)->port) == IB_LINK_LAYER_ETHERNET)
+			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 13;
+		else
+			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 12;
+	}
+
+	return 0;
+}
+
+int mlx4_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)
+{
+	struct mlx4_ib_cq *cq = to_mcq(ibcq);
+	struct mlx4_ib_qp *cur_qp = NULL;
+	unsigned long flags;
+	int npolled;
+	int err = 0;
+
+	spin_lock_irqsave(&cq->lock, flags);
+
+	for (npolled = 0; npolled < num_entries; ++npolled) {
+		err = mlx4_ib_poll_one(cq, &cur_qp, wc + npolled);
+		if (err)
+			break;
+	}
+
+	mlx4_cq_set_ci(&cq->mcq);
+
+	spin_unlock_irqrestore(&cq->lock, flags);
+
+	if (err == 0 || err == -EAGAIN)
+		return npolled;
+	else
+		return err;
+}
+
+int mlx4_ib_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags)
+{
+	mlx4_cq_arm(&to_mcq(ibcq)->mcq,
+		    (flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED ?
+		    MLX4_CQ_DB_REQ_NOT_SOL : MLX4_CQ_DB_REQ_NOT,
+		    to_mdev(ibcq->device)->uar_map,
+		    MLX4_GET_DOORBELL_LOCK(&to_mdev(ibcq->device)->uar_lock));
+
+	return 0;
+}
+
+void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
+{
+	u32 prod_index;
+	int nfreed = 0;
+	struct mlx4_cqe *cqe, *dest;
+	u8 owner_bit;
+	int cqe_inc = cq->buf.entry_size == 64 ? 1 : 0;
+
+	/*
+	 * First we need to find the current producer index, so we
+	 * know where to start cleaning from.  It doesn't matter if HW
+	 * adds new entries after this loop -- the QP we're worried
+	 * about is already in RESET, so the new entries won't come
+	 * from our QP and therefore don't need to be checked.
+	 */
+	for (prod_index = cq->mcq.cons_index; get_sw_cqe(cq, prod_index); ++prod_index)
+		if (prod_index == cq->mcq.cons_index + cq->ibcq.cqe)
+			break;
+
+	/*
+	 * Now sweep backwards through the CQ, removing CQ entries
+	 * that match our QP by copying older entries on top of them.
+	 */
+	while ((int) --prod_index - (int) cq->mcq.cons_index >= 0) {
+		cqe = get_cqe(cq, prod_index & cq->ibcq.cqe);
+		cqe += cqe_inc;
+
+		if ((be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_QPN_MASK) == qpn) {
+			if (srq && !(cqe->owner_sr_opcode & MLX4_CQE_IS_SEND_MASK))
+				mlx4_ib_free_srq_wqe(srq, be16_to_cpu(cqe->wqe_index));
+			++nfreed;
+		} else if (nfreed) {
+			dest = get_cqe(cq, (prod_index + nfreed) & cq->ibcq.cqe);
+			dest += cqe_inc;
+
+			owner_bit = dest->owner_sr_opcode & MLX4_CQE_OWNER_MASK;
+			memcpy(dest, cqe, sizeof *cqe);
+			dest->owner_sr_opcode = owner_bit |
+				(dest->owner_sr_opcode & ~MLX4_CQE_OWNER_MASK);
+		}
+	}
+
+	if (nfreed) {
+		cq->mcq.cons_index += nfreed;
+		/*
+		 * Make sure update of buffer contents is done before
+		 * updating consumer index.
+		 */
+		wmb();
+		mlx4_cq_set_ci(&cq->mcq);
+	}
+}
+
+void mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq)
+{
+	spin_lock_irq(&cq->lock);
+	__mlx4_ib_cq_clean(cq, qpn, srq);
+	spin_unlock_irq(&cq->lock);
+}

drivers/infiniband/hw/mlx4/doorbell.c

@@ -0,0 +1,98 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, 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/slab.h>
+
+#include "mlx4_ib.h"
+
+struct mlx4_ib_user_db_page {
+	struct list_head	list;
+	struct ib_umem	       *umem;
+	unsigned long		user_virt;
+	int			refcnt;
+};
+
+int mlx4_ib_db_map_user(struct mlx4_ib_ucontext *context, unsigned long virt,
+			struct mlx4_db *db)
+{
+	struct mlx4_ib_user_db_page *page;
+	struct ib_umem_chunk *chunk;
+	int err = 0;
+
+	mutex_lock(&context->db_page_mutex);
+
+	list_for_each_entry(page, &context->db_page_list, list)
+		if (page->user_virt == (virt & PAGE_MASK))
+			goto found;
+
+	page = kmalloc(sizeof *page, GFP_KERNEL);
+	if (!page) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	page->user_virt = (virt & PAGE_MASK);
+	page->refcnt    = 0;
+	page->umem      = ib_umem_get(&context->ibucontext, virt & PAGE_MASK,
+				      PAGE_SIZE, 0, 0);
+	if (IS_ERR(page->umem)) {
+		err = PTR_ERR(page->umem);
+		kfree(page);
+		goto out;
+	}
+
+	list_add(&page->list, &context->db_page_list);
+
+found:
+	chunk = list_entry(page->umem->chunk_list.next, struct ib_umem_chunk, list);
+	db->dma		= sg_dma_address(chunk->page_list) + (virt & ~PAGE_MASK);
+	db->u.user_page = page;
+	++page->refcnt;
+
+out:
+	mutex_unlock(&context->db_page_mutex);
+
+	return err;
+}
+
+void mlx4_ib_db_unmap_user(struct mlx4_ib_ucontext *context, struct mlx4_db *db)
+{
+	mutex_lock(&context->db_page_mutex);
+
+	if (!--db->u.user_page->refcnt) {
+		list_del(&db->u.user_page->list);
+		ib_umem_release(db->u.user_page->umem);
+		kfree(db->u.user_page);
+	}
+
+	mutex_unlock(&context->db_page_mutex);
+}

drivers/infiniband/hw/mlx4/mlx4_ib.h

@@ -0,0 +1,744 @@
+/*
+ * Copyright (c) 2006, 2007 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2007, 2008 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.
+ */
+
+#ifndef MLX4_IB_H
+#define MLX4_IB_H
+
+#include <linux/compiler.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/idr.h>
+
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_umem.h>
+#include <rdma/ib_mad.h>
+#include <rdma/ib_sa.h>
+
+#include <linux/mlx4/device.h>
+#include <linux/mlx4/doorbell.h>
+
+#define MLX4_IB_DRV_NAME	"mlx4_ib"
+
+#ifdef pr_fmt
+#undef pr_fmt
+#endif
+#define pr_fmt(fmt)	"<" MLX4_IB_DRV_NAME "> %s: " fmt, __func__
+
+#define mlx4_ib_warn(ibdev, format, arg...) \
+	dev_warn((ibdev)->dma_device, MLX4_IB_DRV_NAME ": " format, ## arg)
+
+enum {
+	MLX4_IB_SQ_MIN_WQE_SHIFT = 6,
+	MLX4_IB_MAX_HEADROOM	 = 2048
+};
+
+#define MLX4_IB_SQ_HEADROOM(shift)	((MLX4_IB_MAX_HEADROOM >> (shift)) + 1)
+#define MLX4_IB_SQ_MAX_SPARE		(MLX4_IB_SQ_HEADROOM(MLX4_IB_SQ_MIN_WQE_SHIFT))
+
+/*module param to indicate if SM assigns the alias_GUID*/
+extern int mlx4_ib_sm_guid_assign;
+
+struct mlx4_ib_ucontext {
+	struct ib_ucontext	ibucontext;
+	struct mlx4_uar		uar;
+	struct list_head	db_page_list;
+	struct mutex		db_page_mutex;
+};
+
+struct mlx4_ib_pd {
+	struct ib_pd		ibpd;
+	u32			pdn;
+};
+
+struct mlx4_ib_xrcd {
+	struct ib_xrcd		ibxrcd;
+	u32			xrcdn;
+	struct ib_pd	       *pd;
+	struct ib_cq	       *cq;
+};
+
+struct mlx4_ib_cq_buf {
+	struct mlx4_buf		buf;
+	struct mlx4_mtt		mtt;
+	int			entry_size;
+};
+
+struct mlx4_ib_cq_resize {
+	struct mlx4_ib_cq_buf	buf;
+	int			cqe;
+};
+
+struct mlx4_ib_cq {
+	struct ib_cq		ibcq;
+	struct mlx4_cq		mcq;
+	struct mlx4_ib_cq_buf	buf;
+	struct mlx4_ib_cq_resize *resize_buf;
+	struct mlx4_db		db;
+	spinlock_t		lock;
+	struct mutex		resize_mutex;
+	struct ib_umem	       *umem;
+	struct ib_umem	       *resize_umem;
+};
+
+struct mlx4_ib_mr {
+	struct ib_mr		ibmr;
+	struct mlx4_mr		mmr;
+	struct ib_umem	       *umem;
+};
+
+struct mlx4_ib_mw {
+	struct ib_mw		ibmw;
+	struct mlx4_mw		mmw;
+};
+
+struct mlx4_ib_fast_reg_page_list {
+	struct ib_fast_reg_page_list	ibfrpl;
+	__be64			       *mapped_page_list;
+	dma_addr_t			map;
+};
+
+struct mlx4_ib_fmr {
+	struct ib_fmr           ibfmr;
+	struct mlx4_fmr         mfmr;
+};
+
+struct mlx4_ib_wq {
+	u64		       *wrid;
+	spinlock_t		lock;
+	int			wqe_cnt;
+	int			max_post;
+	int			max_gs;
+	int			offset;
+	int			wqe_shift;
+	unsigned		head;
+	unsigned		tail;
+};
+
+enum mlx4_ib_qp_flags {
+	MLX4_IB_QP_LSO = IB_QP_CREATE_IPOIB_UD_LSO,
+	MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK = IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK,
+	MLX4_IB_SRIOV_TUNNEL_QP = 1 << 30,
+	MLX4_IB_SRIOV_SQP = 1 << 31,
+};
+
+struct mlx4_ib_gid_entry {
+	struct list_head	list;
+	union ib_gid		gid;
+	int			added;
+	u8			port;
+};
+
+enum mlx4_ib_qp_type {
+	/*
+	 * IB_QPT_SMI and IB_QPT_GSI have to be the first two entries
+	 * here (and in that order) since the MAD layer uses them as
+	 * indices into a 2-entry table.
+	 */
+	MLX4_IB_QPT_SMI = IB_QPT_SMI,
+	MLX4_IB_QPT_GSI = IB_QPT_GSI,
+
+	MLX4_IB_QPT_RC = IB_QPT_RC,
+	MLX4_IB_QPT_UC = IB_QPT_UC,
+	MLX4_IB_QPT_UD = IB_QPT_UD,
+	MLX4_IB_QPT_RAW_IPV6 = IB_QPT_RAW_IPV6,
+	MLX4_IB_QPT_RAW_ETHERTYPE = IB_QPT_RAW_ETHERTYPE,
+	MLX4_IB_QPT_RAW_PACKET = IB_QPT_RAW_PACKET,
+	MLX4_IB_QPT_XRC_INI = IB_QPT_XRC_INI,
+	MLX4_IB_QPT_XRC_TGT = IB_QPT_XRC_TGT,
+
+	MLX4_IB_QPT_PROXY_SMI_OWNER	= 1 << 16,
+	MLX4_IB_QPT_PROXY_SMI		= 1 << 17,
+	MLX4_IB_QPT_PROXY_GSI		= 1 << 18,
+	MLX4_IB_QPT_TUN_SMI_OWNER	= 1 << 19,
+	MLX4_IB_QPT_TUN_SMI		= 1 << 20,
+	MLX4_IB_QPT_TUN_GSI		= 1 << 21,
+};
+
+#define MLX4_IB_QPT_ANY_SRIOV	(MLX4_IB_QPT_PROXY_SMI_OWNER | \
+	MLX4_IB_QPT_PROXY_SMI | MLX4_IB_QPT_PROXY_GSI | MLX4_IB_QPT_TUN_SMI_OWNER | \
+	MLX4_IB_QPT_TUN_SMI | MLX4_IB_QPT_TUN_GSI)
+
+enum mlx4_ib_mad_ifc_flags {
+	MLX4_MAD_IFC_IGNORE_MKEY	= 1,
+	MLX4_MAD_IFC_IGNORE_BKEY	= 2,
+	MLX4_MAD_IFC_IGNORE_KEYS	= (MLX4_MAD_IFC_IGNORE_MKEY |
+					   MLX4_MAD_IFC_IGNORE_BKEY),
+	MLX4_MAD_IFC_NET_VIEW		= 4,
+};
+
+enum {
+	MLX4_NUM_TUNNEL_BUFS		= 256,
+};
+
+struct mlx4_ib_tunnel_header {
+	struct mlx4_av av;
+	__be32 remote_qpn;
+	__be32 qkey;
+	__be16 vlan;
+	u8 mac[6];
+	__be16 pkey_index;
+	u8 reserved[6];
+};
+
+struct mlx4_ib_buf {
+	void *addr;
+	dma_addr_t map;
+};
+
+struct mlx4_rcv_tunnel_hdr {
+	__be32 flags_src_qp; /* flags[6:5] is defined for VLANs:
+			      * 0x0 - no vlan was in the packet
+			      * 0x01 - C-VLAN was in the packet */
+	u8 g_ml_path; /* gid bit stands for ipv6/4 header in RoCE */
+	u8 reserved;
+	__be16 pkey_index;
+	__be16 sl_vid;
+	__be16 slid_mac_47_32;
+	__be32 mac_31_0;
+};
+
+struct mlx4_ib_proxy_sqp_hdr {
+	struct ib_grh grh;
+	struct mlx4_rcv_tunnel_hdr tun;
+}  __packed;
+
+struct mlx4_ib_qp {
+	struct ib_qp		ibqp;
+	struct mlx4_qp		mqp;
+	struct mlx4_buf		buf;
+
+	struct mlx4_db		db;
+	struct mlx4_ib_wq	rq;
+
+	u32			doorbell_qpn;
+	__be32			sq_signal_bits;
+	unsigned		sq_next_wqe;
+	int			sq_max_wqes_per_wr;
+	int			sq_spare_wqes;
+	struct mlx4_ib_wq	sq;
+
+	enum mlx4_ib_qp_type	mlx4_ib_qp_type;
+	struct ib_umem	       *umem;
+	struct mlx4_mtt		mtt;
+	int			buf_size;
+	struct mutex		mutex;
+	u16			xrcdn;
+	u32			flags;
+	u8			port;
+	u8			alt_port;
+	u8			atomic_rd_en;
+	u8			resp_depth;
+	u8			sq_no_prefetch;
+	u8			state;
+	int			mlx_type;
+	struct list_head	gid_list;
+	struct list_head	steering_rules;
+	struct mlx4_ib_buf	*sqp_proxy_rcv;
+
+};
+
+struct mlx4_ib_srq {
+	struct ib_srq		ibsrq;
+	struct mlx4_srq		msrq;
+	struct mlx4_buf		buf;
+	struct mlx4_db		db;
+	u64		       *wrid;
+	spinlock_t		lock;
+	int			head;
+	int			tail;
+	u16			wqe_ctr;
+	struct ib_umem	       *umem;
+	struct mlx4_mtt		mtt;
+	struct mutex		mutex;
+};
+
+struct mlx4_ib_ah {
+	struct ib_ah		ibah;
+	union mlx4_ext_av       av;
+};
+
+/****************************************/
+/* alias guid support */
+/****************************************/
+#define NUM_PORT_ALIAS_GUID		2
+#define NUM_ALIAS_GUID_IN_REC		8
+#define NUM_ALIAS_GUID_REC_IN_PORT	16
+#define GUID_REC_SIZE			8
+#define NUM_ALIAS_GUID_PER_PORT		128
+#define MLX4_NOT_SET_GUID		(0x00LL)
+#define MLX4_GUID_FOR_DELETE_VAL	(~(0x00LL))
+
+enum mlx4_guid_alias_rec_status {
+	MLX4_GUID_INFO_STATUS_IDLE,
+	MLX4_GUID_INFO_STATUS_SET,
+	MLX4_GUID_INFO_STATUS_PENDING,
+};
+
+enum mlx4_guid_alias_rec_ownership {
+	MLX4_GUID_DRIVER_ASSIGN,
+	MLX4_GUID_SYSADMIN_ASSIGN,
+	MLX4_GUID_NONE_ASSIGN, /*init state of each record*/
+};
+
+enum mlx4_guid_alias_rec_method {
+	MLX4_GUID_INFO_RECORD_SET	= IB_MGMT_METHOD_SET,
+	MLX4_GUID_INFO_RECORD_DELETE	= IB_SA_METHOD_DELETE,
+};
+
+struct mlx4_sriov_alias_guid_info_rec_det {
+	u8 all_recs[GUID_REC_SIZE * NUM_ALIAS_GUID_IN_REC];
+	ib_sa_comp_mask guid_indexes; /*indicates what from the 8 records are valid*/
+	enum mlx4_guid_alias_rec_status status; /*indicates the administraively status of the record.*/
+	u8 method; /*set or delete*/
+	enum mlx4_guid_alias_rec_ownership ownership; /*indicates who assign that alias_guid record*/
+};
+
+struct mlx4_sriov_alias_guid_port_rec_det {
+	struct mlx4_sriov_alias_guid_info_rec_det all_rec_per_port[NUM_ALIAS_GUID_REC_IN_PORT];
+	struct workqueue_struct *wq;
+	struct delayed_work alias_guid_work;
+	u8 port;
+	struct mlx4_sriov_alias_guid *parent;
+	struct list_head cb_list;
+};
+
+struct mlx4_sriov_alias_guid {
+	struct mlx4_sriov_alias_guid_port_rec_det ports_guid[MLX4_MAX_PORTS];
+	spinlock_t ag_work_lock;
+	struct ib_sa_client *sa_client;
+};
+
+struct mlx4_ib_demux_work {
+	struct work_struct	work;
+	struct mlx4_ib_dev     *dev;
+	int			slave;
+	int			do_init;
+	u8			port;
+
+};
+
+struct mlx4_ib_tun_tx_buf {
+	struct mlx4_ib_buf buf;
+	struct ib_ah *ah;
+};
+
+struct mlx4_ib_demux_pv_qp {
+	struct ib_qp *qp;
+	enum ib_qp_type proxy_qpt;
+	struct mlx4_ib_buf *ring;
+	struct mlx4_ib_tun_tx_buf *tx_ring;
+	spinlock_t tx_lock;
+	unsigned tx_ix_head;
+	unsigned tx_ix_tail;
+};
+
+enum mlx4_ib_demux_pv_state {
+	DEMUX_PV_STATE_DOWN,
+	DEMUX_PV_STATE_STARTING,
+	DEMUX_PV_STATE_ACTIVE,
+	DEMUX_PV_STATE_DOWNING,
+};
+
+struct mlx4_ib_demux_pv_ctx {
+	int port;
+	int slave;
+	enum mlx4_ib_demux_pv_state state;
+	int has_smi;
+	struct ib_device *ib_dev;
+	struct ib_cq *cq;
+	struct ib_pd *pd;
+	struct ib_mr *mr;
+	struct work_struct work;
+	struct workqueue_struct *wq;
+	struct mlx4_ib_demux_pv_qp qp[2];
+};
+
+struct mlx4_ib_demux_ctx {
+	struct ib_device *ib_dev;
+	int port;
+	struct workqueue_struct *wq;
+	struct workqueue_struct *ud_wq;
+	spinlock_t ud_lock;
+	__be64 subnet_prefix;
+	__be64 guid_cache[128];
+	struct mlx4_ib_dev *dev;
+	/* the following lock protects both mcg_table and mcg_mgid0_list */
+	struct mutex		mcg_table_lock;
+	struct rb_root		mcg_table;
+	struct list_head	mcg_mgid0_list;
+	struct workqueue_struct	*mcg_wq;
+	struct mlx4_ib_demux_pv_ctx **tun;
+	atomic_t tid;
+	int    flushing; /* flushing the work queue */
+};
+
+struct mlx4_ib_sriov {
+	struct mlx4_ib_demux_ctx demux[MLX4_MAX_PORTS];
+	struct mlx4_ib_demux_pv_ctx *sqps[MLX4_MAX_PORTS];
+	/* when using this spinlock you should use "irq" because
+	 * it may be called from interrupt context.*/
+	spinlock_t going_down_lock;
+	int is_going_down;
+
+	struct mlx4_sriov_alias_guid alias_guid;
+
+	/* CM paravirtualization fields */
+	struct list_head cm_list;
+	spinlock_t id_map_lock;
+	struct rb_root sl_id_map;
+	struct idr pv_id_table;
+};
+
+struct mlx4_ib_iboe {
+	spinlock_t		lock;
+	struct net_device      *netdevs[MLX4_MAX_PORTS];
+	struct notifier_block 	nb;
+	union ib_gid		gid_table[MLX4_MAX_PORTS][128];
+};
+
+struct pkey_mgt {
+	u8			virt2phys_pkey[MLX4_MFUNC_MAX][MLX4_MAX_PORTS][MLX4_MAX_PORT_PKEYS];
+	u16			phys_pkey_cache[MLX4_MAX_PORTS][MLX4_MAX_PORT_PKEYS];
+	struct list_head	pkey_port_list[MLX4_MFUNC_MAX];
+	struct kobject	       *device_parent[MLX4_MFUNC_MAX];
+};
+
+struct mlx4_ib_iov_sysfs_attr {
+	void *ctx;
+	struct kobject *kobj;
+	unsigned long data;
+	u32 entry_num;
+	char name[15];
+	struct device_attribute dentry;
+	struct device *dev;
+};
+
+struct mlx4_ib_iov_sysfs_attr_ar {
+	struct mlx4_ib_iov_sysfs_attr dentries[3 * NUM_ALIAS_GUID_PER_PORT + 1];
+};
+
+struct mlx4_ib_iov_port {
+	char name[100];
+	u8 num;
+	struct mlx4_ib_dev *dev;
+	struct list_head list;
+	struct mlx4_ib_iov_sysfs_attr_ar *dentr_ar;
+	struct ib_port_attr attr;
+	struct kobject	*cur_port;
+	struct kobject	*admin_alias_parent;
+	struct kobject	*gids_parent;
+	struct kobject	*pkeys_parent;
+	struct kobject	*mcgs_parent;
+	struct mlx4_ib_iov_sysfs_attr mcg_dentry;
+};
+
+struct mlx4_ib_dev {
+	struct ib_device	ib_dev;
+	struct mlx4_dev	       *dev;
+	int			num_ports;
+	void __iomem	       *uar_map;
+
+	struct mlx4_uar		priv_uar;
+	u32			priv_pdn;
+	MLX4_DECLARE_DOORBELL_LOCK(uar_lock);
+
+	struct ib_mad_agent    *send_agent[MLX4_MAX_PORTS][2];
+	struct ib_ah	       *sm_ah[MLX4_MAX_PORTS];
+	spinlock_t		sm_lock;
+	struct mlx4_ib_sriov	sriov;
+
+	struct mutex		cap_mask_mutex;
+	bool			ib_active;
+	struct mlx4_ib_iboe	iboe;
+	int			counters[MLX4_MAX_PORTS];
+	int		       *eq_table;
+	int			eq_added;
+	struct kobject	       *iov_parent;
+	struct kobject	       *ports_parent;
+	struct kobject	       *dev_ports_parent[MLX4_MFUNC_MAX];
+	struct mlx4_ib_iov_port	iov_ports[MLX4_MAX_PORTS];
+	struct pkey_mgt		pkeys;
+};
+
+struct ib_event_work {
+	struct work_struct	work;
+	struct mlx4_ib_dev	*ib_dev;
+	struct mlx4_eqe		ib_eqe;
+};
+
+struct mlx4_ib_qp_tunnel_init_attr {
+	struct ib_qp_init_attr init_attr;
+	int slave;
+	enum ib_qp_type proxy_qp_type;
+	u8 port;
+};
+
+static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev)
+{
+	return container_of(ibdev, struct mlx4_ib_dev, ib_dev);
+}
+
+static inline struct mlx4_ib_ucontext *to_mucontext(struct ib_ucontext *ibucontext)
+{
+	return container_of(ibucontext, struct mlx4_ib_ucontext, ibucontext);
+}
+
+static inline struct mlx4_ib_pd *to_mpd(struct ib_pd *ibpd)
+{
+	return container_of(ibpd, struct mlx4_ib_pd, ibpd);
+}
+
+static inline struct mlx4_ib_xrcd *to_mxrcd(struct ib_xrcd *ibxrcd)
+{
+	return container_of(ibxrcd, struct mlx4_ib_xrcd, ibxrcd);
+}
+
+static inline struct mlx4_ib_cq *to_mcq(struct ib_cq *ibcq)
+{
+	return container_of(ibcq, struct mlx4_ib_cq, ibcq);
+}
+
+static inline struct mlx4_ib_cq *to_mibcq(struct mlx4_cq *mcq)
+{
+	return container_of(mcq, struct mlx4_ib_cq, mcq);
+}
+
+static inline struct mlx4_ib_mr *to_mmr(struct ib_mr *ibmr)
+{
+	return container_of(ibmr, struct mlx4_ib_mr, ibmr);
+}
+
+static inline struct mlx4_ib_mw *to_mmw(struct ib_mw *ibmw)
+{
+	return container_of(ibmw, struct mlx4_ib_mw, ibmw);
+}
+
+static inline struct mlx4_ib_fast_reg_page_list *to_mfrpl(struct ib_fast_reg_page_list *ibfrpl)
+{
+	return container_of(ibfrpl, struct mlx4_ib_fast_reg_page_list, ibfrpl);
+}
+
+static inline struct mlx4_ib_fmr *to_mfmr(struct ib_fmr *ibfmr)
+{
+	return container_of(ibfmr, struct mlx4_ib_fmr, ibfmr);
+}
+static inline struct mlx4_ib_qp *to_mqp(struct ib_qp *ibqp)
+{
+	return container_of(ibqp, struct mlx4_ib_qp, ibqp);
+}
+
+static inline struct mlx4_ib_qp *to_mibqp(struct mlx4_qp *mqp)
+{
+	return container_of(mqp, struct mlx4_ib_qp, mqp);
+}
+
+static inline struct mlx4_ib_srq *to_msrq(struct ib_srq *ibsrq)
+{
+	return container_of(ibsrq, struct mlx4_ib_srq, ibsrq);
+}
+
+static inline struct mlx4_ib_srq *to_mibsrq(struct mlx4_srq *msrq)
+{
+	return container_of(msrq, struct mlx4_ib_srq, msrq);
+}
+
+static inline struct mlx4_ib_ah *to_mah(struct ib_ah *ibah)
+{
+	return container_of(ibah, struct mlx4_ib_ah, ibah);
+}
+
+int mlx4_ib_init_sriov(struct mlx4_ib_dev *dev);
+void mlx4_ib_close_sriov(struct mlx4_ib_dev *dev);
+
+int mlx4_ib_db_map_user(struct mlx4_ib_ucontext *context, unsigned long virt,
+			struct mlx4_db *db);
+void mlx4_ib_db_unmap_user(struct mlx4_ib_ucontext *context, struct mlx4_db *db);
+
+struct ib_mr *mlx4_ib_get_dma_mr(struct ib_pd *pd, int acc);
+int mlx4_ib_umem_write_mtt(struct mlx4_ib_dev *dev, struct mlx4_mtt *mtt,
+			   struct ib_umem *umem);
+struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				  u64 virt_addr, int access_flags,
+				  struct ib_udata *udata);
+int mlx4_ib_dereg_mr(struct ib_mr *mr);
+struct ib_mw *mlx4_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type);
+int mlx4_ib_bind_mw(struct ib_qp *qp, struct ib_mw *mw,
+		    struct ib_mw_bind *mw_bind);
+int mlx4_ib_dealloc_mw(struct ib_mw *mw);
+struct ib_mr *mlx4_ib_alloc_fast_reg_mr(struct ib_pd *pd,
+					int max_page_list_len);
+struct ib_fast_reg_page_list *mlx4_ib_alloc_fast_reg_page_list(struct ib_device *ibdev,
+							       int page_list_len);
+void mlx4_ib_free_fast_reg_page_list(struct ib_fast_reg_page_list *page_list);
+
+int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period);
+int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata);
+struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
+				struct ib_ucontext *context,
+				struct ib_udata *udata);
+int mlx4_ib_destroy_cq(struct ib_cq *cq);
+int mlx4_ib_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc);
+int mlx4_ib_arm_cq(struct ib_cq *cq, enum ib_cq_notify_flags flags);
+void __mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq);
+void mlx4_ib_cq_clean(struct mlx4_ib_cq *cq, u32 qpn, struct mlx4_ib_srq *srq);
+
+struct ib_ah *mlx4_ib_create_ah(struct ib_pd *pd, struct ib_ah_attr *ah_attr);
+int mlx4_ib_query_ah(struct ib_ah *ibah, struct ib_ah_attr *ah_attr);
+int mlx4_ib_destroy_ah(struct ib_ah *ah);
+
+struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd,
+				  struct ib_srq_init_attr *init_attr,
+				  struct ib_udata *udata);
+int mlx4_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr,
+		       enum ib_srq_attr_mask attr_mask, struct ib_udata *udata);
+int mlx4_ib_query_srq(struct ib_srq *srq, struct ib_srq_attr *srq_attr);
+int mlx4_ib_destroy_srq(struct ib_srq *srq);
+void mlx4_ib_free_srq_wqe(struct mlx4_ib_srq *srq, int wqe_index);
+int mlx4_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
+			  struct ib_recv_wr **bad_wr);
+
+struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
+				struct ib_qp_init_attr *init_attr,
+				struct ib_udata *udata);
+int mlx4_ib_destroy_qp(struct ib_qp *qp);
+int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		      int attr_mask, struct ib_udata *udata);
+int mlx4_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_mask,
+		     struct ib_qp_init_attr *qp_init_attr);
+int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
+		      struct ib_send_wr **bad_wr);
+int mlx4_ib_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
+		      struct ib_recv_wr **bad_wr);
+
+int mlx4_MAD_IFC(struct mlx4_ib_dev *dev, int mad_ifc_flags,
+		 int port, struct ib_wc *in_wc, struct ib_grh *in_grh,
+		 void *in_mad, void *response_mad);
+int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
+			struct ib_wc *in_wc, struct ib_grh *in_grh,
+			struct ib_mad *in_mad, struct ib_mad *out_mad);
+int mlx4_ib_mad_init(struct mlx4_ib_dev *dev);
+void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev);
+
+struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int mr_access_flags,
+				  struct ib_fmr_attr *fmr_attr);
+int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, int npages,
+			 u64 iova);
+int mlx4_ib_unmap_fmr(struct list_head *fmr_list);
+int mlx4_ib_fmr_dealloc(struct ib_fmr *fmr);
+int __mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
+			 struct ib_port_attr *props, int netw_view);
+int __mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
+			 u16 *pkey, int netw_view);
+
+int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
+			union ib_gid *gid, int netw_view);
+
+int mlx4_ib_resolve_grh(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah_attr,
+			u8 *mac, int *is_mcast, u8 port);
+
+static inline bool mlx4_ib_ah_grh_present(struct mlx4_ib_ah *ah)
+{
+	u8 port = be32_to_cpu(ah->av.ib.port_pd) >> 24 & 3;
+
+	if (rdma_port_get_link_layer(ah->ibah.device, port) == IB_LINK_LAYER_ETHERNET)
+		return true;
+
+	return !!(ah->av.ib.g_slid & 0x80);
+}
+
+int mlx4_ib_mcg_port_init(struct mlx4_ib_demux_ctx *ctx);
+void mlx4_ib_mcg_port_cleanup(struct mlx4_ib_demux_ctx *ctx, int destroy_wq);
+void clean_vf_mcast(struct mlx4_ib_demux_ctx *ctx, int slave);
+int mlx4_ib_mcg_init(void);
+void mlx4_ib_mcg_destroy(void);
+
+int mlx4_ib_find_real_gid(struct ib_device *ibdev, u8 port, __be64 guid);
+
+int mlx4_ib_mcg_multiplex_handler(struct ib_device *ibdev, int port, int slave,
+				  struct ib_sa_mad *sa_mad);
+int mlx4_ib_mcg_demux_handler(struct ib_device *ibdev, int port, int slave,
+			      struct ib_sa_mad *mad);
+
+int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp,
+		   union ib_gid *gid);
+
+void mlx4_ib_dispatch_event(struct mlx4_ib_dev *dev, u8 port_num,
+			    enum ib_event_type type);
+
+void mlx4_ib_tunnels_update_work(struct work_struct *work);
+
+int mlx4_ib_send_to_slave(struct mlx4_ib_dev *dev, int slave, u8 port,
+			  enum ib_qp_type qpt, struct ib_wc *wc,
+			  struct ib_grh *grh, struct ib_mad *mad);
+int mlx4_ib_send_to_wire(struct mlx4_ib_dev *dev, int slave, u8 port,
+			 enum ib_qp_type dest_qpt, u16 pkey_index, u32 remote_qpn,
+			 u32 qkey, struct ib_ah_attr *attr, struct ib_mad *mad);
+__be64 mlx4_ib_get_new_demux_tid(struct mlx4_ib_demux_ctx *ctx);
+
+int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
+		struct ib_mad *mad);
+
+int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id,
+		struct ib_mad *mad);
+
+void mlx4_ib_cm_paravirt_init(struct mlx4_ib_dev *dev);
+void mlx4_ib_cm_paravirt_clean(struct mlx4_ib_dev *dev, int slave_id);
+
+/* alias guid support */
+void mlx4_ib_init_alias_guid_work(struct mlx4_ib_dev *dev, int port);
+int mlx4_ib_init_alias_guid_service(struct mlx4_ib_dev *dev);
+void mlx4_ib_destroy_alias_guid_service(struct mlx4_ib_dev *dev);
+void mlx4_ib_invalidate_all_guid_record(struct mlx4_ib_dev *dev, int port);
+
+void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev,
+					  int block_num,
+					  u8 port_num, u8 *p_data);
+
+void mlx4_ib_update_cache_on_guid_change(struct mlx4_ib_dev *dev,
+					 int block_num, u8 port_num,
+					 u8 *p_data);
+
+int add_sysfs_port_mcg_attr(struct mlx4_ib_dev *device, int port_num,
+			    struct attribute *attr);
+void del_sysfs_port_mcg_attr(struct mlx4_ib_dev *device, int port_num,
+			     struct attribute *attr);
+ib_sa_comp_mask mlx4_ib_get_aguid_comp_mask_from_ix(int index);
+
+int mlx4_ib_device_register_sysfs(struct mlx4_ib_dev *device) ;
+
+void mlx4_ib_device_unregister_sysfs(struct mlx4_ib_dev *device);
+
+__be64 mlx4_ib_gen_node_guid(void);
+
+
+#endif /* MLX4_IB_H */

drivers/infiniband/hw/mlx4/mr.c

@@ -0,0 +1,436 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 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/slab.h>
+
+#include "mlx4_ib.h"
+
+static u32 convert_access(int acc)
+{
+	return (acc & IB_ACCESS_REMOTE_ATOMIC ? MLX4_PERM_ATOMIC       : 0) |
+	       (acc & IB_ACCESS_REMOTE_WRITE  ? MLX4_PERM_REMOTE_WRITE : 0) |
+	       (acc & IB_ACCESS_REMOTE_READ   ? MLX4_PERM_REMOTE_READ  : 0) |
+	       (acc & IB_ACCESS_LOCAL_WRITE   ? MLX4_PERM_LOCAL_WRITE  : 0) |
+	       (acc & IB_ACCESS_MW_BIND	      ? MLX4_PERM_BIND_MW      : 0) |
+	       MLX4_PERM_LOCAL_READ;
+}
+
+static enum mlx4_mw_type to_mlx4_type(enum ib_mw_type type)
+{
+	switch (type) {
+	case IB_MW_TYPE_1:	return MLX4_MW_TYPE_1;
+	case IB_MW_TYPE_2:	return MLX4_MW_TYPE_2;
+	default:		return -1;
+	}
+}
+
+struct ib_mr *mlx4_ib_get_dma_mr(struct ib_pd *pd, int acc)
+{
+	struct mlx4_ib_mr *mr;
+	int err;
+
+	mr = kmalloc(sizeof *mr, GFP_KERNEL);
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_mr_alloc(to_mdev(pd->device)->dev, to_mpd(pd)->pdn, 0,
+			    ~0ull, convert_access(acc), 0, 0, &mr->mmr);
+	if (err)
+		goto err_free;
+
+	err = mlx4_mr_enable(to_mdev(pd->device)->dev, &mr->mmr);
+	if (err)
+		goto err_mr;
+
+	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+	mr->umem = NULL;
+
+	return &mr->ibmr;
+
+err_mr:
+	(void) mlx4_mr_free(to_mdev(pd->device)->dev, &mr->mmr);
+
+err_free:
+	kfree(mr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_umem_write_mtt(struct mlx4_ib_dev *dev, struct mlx4_mtt *mtt,
+			   struct ib_umem *umem)
+{
+	u64 *pages;
+	struct ib_umem_chunk *chunk;
+	int i, j, k;
+	int n;
+	int len;
+	int err = 0;
+
+	pages = (u64 *) __get_free_page(GFP_KERNEL);
+	if (!pages)
+		return -ENOMEM;
+
+	i = n = 0;
+
+	list_for_each_entry(chunk, &umem->chunk_list, list)
+		for (j = 0; j < chunk->nmap; ++j) {
+			len = sg_dma_len(&chunk->page_list[j]) >> mtt->page_shift;
+			for (k = 0; k < len; ++k) {
+				pages[i++] = sg_dma_address(&chunk->page_list[j]) +
+					umem->page_size * k;
+				/*
+				 * Be friendly to mlx4_write_mtt() and
+				 * pass it chunks of appropriate size.
+				 */
+				if (i == PAGE_SIZE / sizeof (u64)) {
+					err = mlx4_write_mtt(dev->dev, mtt, n,
+							     i, pages);
+					if (err)
+						goto out;
+					n += i;
+					i = 0;
+				}
+			}
+		}
+
+	if (i)
+		err = mlx4_write_mtt(dev->dev, mtt, n, i, pages);
+
+out:
+	free_page((unsigned long) pages);
+	return err;
+}
+
+struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
+				  u64 virt_addr, int access_flags,
+				  struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_mr *mr;
+	int shift;
+	int err;
+	int n;
+
+	mr = kmalloc(sizeof *mr, GFP_KERNEL);
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
+
+	mr->umem = ib_umem_get(pd->uobject->context, start, length,
+			       access_flags, 0);
+	if (IS_ERR(mr->umem)) {
+		err = PTR_ERR(mr->umem);
+		goto err_free;
+	}
+
+	n = ib_umem_page_count(mr->umem);
+	shift = ilog2(mr->umem->page_size);
+
+	err = mlx4_mr_alloc(dev->dev, to_mpd(pd)->pdn, virt_addr, length,
+			    convert_access(access_flags), n, shift, &mr->mmr);
+	if (err)
+		goto err_umem;
+
+	err = mlx4_ib_umem_write_mtt(dev, &mr->mmr.mtt, mr->umem);
+	if (err)
+		goto err_mr;
+
+	err = mlx4_mr_enable(dev->dev, &mr->mmr);
+	if (err)
+		goto err_mr;
+
+	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+
+	return &mr->ibmr;
+
+err_mr:
+	(void) mlx4_mr_free(to_mdev(pd->device)->dev, &mr->mmr);
+
+err_umem:
+	ib_umem_release(mr->umem);
+
+err_free:
+	kfree(mr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_dereg_mr(struct ib_mr *ibmr)
+{
+	struct mlx4_ib_mr *mr = to_mmr(ibmr);
+	int ret;
+
+	ret = mlx4_mr_free(to_mdev(ibmr->device)->dev, &mr->mmr);
+	if (ret)
+		return ret;
+	if (mr->umem)
+		ib_umem_release(mr->umem);
+	kfree(mr);
+
+	return 0;
+}
+
+struct ib_mw *mlx4_ib_alloc_mw(struct ib_pd *pd, enum ib_mw_type type)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_mw *mw;
+	int err;
+
+	mw = kmalloc(sizeof(*mw), GFP_KERNEL);
+	if (!mw)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_mw_alloc(dev->dev, to_mpd(pd)->pdn,
+			    to_mlx4_type(type), &mw->mmw);
+	if (err)
+		goto err_free;
+
+	err = mlx4_mw_enable(dev->dev, &mw->mmw);
+	if (err)
+		goto err_mw;
+
+	mw->ibmw.rkey = mw->mmw.key;
+
+	return &mw->ibmw;
+
+err_mw:
+	mlx4_mw_free(dev->dev, &mw->mmw);
+
+err_free:
+	kfree(mw);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_bind_mw(struct ib_qp *qp, struct ib_mw *mw,
+		    struct ib_mw_bind *mw_bind)
+{
+	struct ib_send_wr  wr;
+	struct ib_send_wr *bad_wr;
+	int ret;
+
+	memset(&wr, 0, sizeof(wr));
+	wr.opcode               = IB_WR_BIND_MW;
+	wr.wr_id                = mw_bind->wr_id;
+	wr.send_flags           = mw_bind->send_flags;
+	wr.wr.bind_mw.mw        = mw;
+	wr.wr.bind_mw.bind_info = mw_bind->bind_info;
+	wr.wr.bind_mw.rkey      = ib_inc_rkey(mw->rkey);
+
+	ret = mlx4_ib_post_send(qp, &wr, &bad_wr);
+	if (!ret)
+		mw->rkey = wr.wr.bind_mw.rkey;
+
+	return ret;
+}
+
+int mlx4_ib_dealloc_mw(struct ib_mw *ibmw)
+{
+	struct mlx4_ib_mw *mw = to_mmw(ibmw);
+
+	mlx4_mw_free(to_mdev(ibmw->device)->dev, &mw->mmw);
+	kfree(mw);
+
+	return 0;
+}
+
+struct ib_mr *mlx4_ib_alloc_fast_reg_mr(struct ib_pd *pd,
+					int max_page_list_len)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_mr *mr;
+	int err;
+
+	mr = kmalloc(sizeof *mr, GFP_KERNEL);
+	if (!mr)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_mr_alloc(dev->dev, to_mpd(pd)->pdn, 0, 0, 0,
+			    max_page_list_len, 0, &mr->mmr);
+	if (err)
+		goto err_free;
+
+	err = mlx4_mr_enable(dev->dev, &mr->mmr);
+	if (err)
+		goto err_mr;
+
+	mr->ibmr.rkey = mr->ibmr.lkey = mr->mmr.key;
+	mr->umem = NULL;
+
+	return &mr->ibmr;
+
+err_mr:
+	(void) mlx4_mr_free(dev->dev, &mr->mmr);
+
+err_free:
+	kfree(mr);
+	return ERR_PTR(err);
+}
+
+struct ib_fast_reg_page_list *mlx4_ib_alloc_fast_reg_page_list(struct ib_device *ibdev,
+							       int page_list_len)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibdev);
+	struct mlx4_ib_fast_reg_page_list *mfrpl;
+	int size = page_list_len * sizeof (u64);
+
+	if (page_list_len > MLX4_MAX_FAST_REG_PAGES)
+		return ERR_PTR(-EINVAL);
+
+	mfrpl = kmalloc(sizeof *mfrpl, GFP_KERNEL);
+	if (!mfrpl)
+		return ERR_PTR(-ENOMEM);
+
+	mfrpl->ibfrpl.page_list = kmalloc(size, GFP_KERNEL);
+	if (!mfrpl->ibfrpl.page_list)
+		goto err_free;
+
+	mfrpl->mapped_page_list = dma_alloc_coherent(&dev->dev->pdev->dev,
+						     size, &mfrpl->map,
+						     GFP_KERNEL);
+	if (!mfrpl->mapped_page_list)
+		goto err_free;
+
+	WARN_ON(mfrpl->map & 0x3f);
+
+	return &mfrpl->ibfrpl;
+
+err_free:
+	kfree(mfrpl->ibfrpl.page_list);
+	kfree(mfrpl);
+	return ERR_PTR(-ENOMEM);
+}
+
+void mlx4_ib_free_fast_reg_page_list(struct ib_fast_reg_page_list *page_list)
+{
+	struct mlx4_ib_dev *dev = to_mdev(page_list->device);
+	struct mlx4_ib_fast_reg_page_list *mfrpl = to_mfrpl(page_list);
+	int size = page_list->max_page_list_len * sizeof (u64);
+
+	dma_free_coherent(&dev->dev->pdev->dev, size, mfrpl->mapped_page_list,
+			  mfrpl->map);
+	kfree(mfrpl->ibfrpl.page_list);
+	kfree(mfrpl);
+}
+
+struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int acc,
+				 struct ib_fmr_attr *fmr_attr)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_fmr *fmr;
+	int err = -ENOMEM;
+
+	fmr = kmalloc(sizeof *fmr, GFP_KERNEL);
+	if (!fmr)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_fmr_alloc(dev->dev, to_mpd(pd)->pdn, convert_access(acc),
+			     fmr_attr->max_pages, fmr_attr->max_maps,
+			     fmr_attr->page_shift, &fmr->mfmr);
+	if (err)
+		goto err_free;
+
+	err = mlx4_fmr_enable(to_mdev(pd->device)->dev, &fmr->mfmr);
+	if (err)
+		goto err_mr;
+
+	fmr->ibfmr.rkey = fmr->ibfmr.lkey = fmr->mfmr.mr.key;
+
+	return &fmr->ibfmr;
+
+err_mr:
+	(void) mlx4_mr_free(to_mdev(pd->device)->dev, &fmr->mfmr.mr);
+
+err_free:
+	kfree(fmr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list,
+		      int npages, u64 iova)
+{
+	struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+	struct mlx4_ib_dev *dev = to_mdev(ifmr->ibfmr.device);
+
+	return mlx4_map_phys_fmr(dev->dev, &ifmr->mfmr, page_list, npages, iova,
+				 &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey);
+}
+
+int mlx4_ib_unmap_fmr(struct list_head *fmr_list)
+{
+	struct ib_fmr *ibfmr;
+	int err;
+	struct mlx4_dev *mdev = NULL;
+
+	list_for_each_entry(ibfmr, fmr_list, list) {
+		if (mdev && to_mdev(ibfmr->device)->dev != mdev)
+			return -EINVAL;
+		mdev = to_mdev(ibfmr->device)->dev;
+	}
+
+	if (!mdev)
+		return 0;
+
+	list_for_each_entry(ibfmr, fmr_list, list) {
+		struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+
+		mlx4_fmr_unmap(mdev, &ifmr->mfmr, &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey);
+	}
+
+	/*
+	 * Make sure all MPT status updates are visible before issuing
+	 * SYNC_TPT firmware command.
+	 */
+	wmb();
+
+	err = mlx4_SYNC_TPT(mdev);
+	if (err)
+		pr_warn("SYNC_TPT error %d when "
+		       "unmapping FMRs\n", err);
+
+	return 0;
+}
+
+int mlx4_ib_fmr_dealloc(struct ib_fmr *ibfmr)
+{
+	struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+	struct mlx4_ib_dev *dev = to_mdev(ibfmr->device);
+	int err;
+
+	err = mlx4_fmr_free(dev->dev, &ifmr->mfmr);
+
+	if (!err)
+		kfree(ifmr);
+
+	return err;
+}

drivers/infiniband/hw/mlx4/srq.c

@@ -0,0 +1,368 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 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/mlx4/qp.h>
+#include <linux/mlx4/srq.h>
+#include <linux/slab.h>
+
+#include "mlx4_ib.h"
+#include "user.h"
+
+static void *get_wqe(struct mlx4_ib_srq *srq, int n)
+{
+	return mlx4_buf_offset(&srq->buf, n << srq->msrq.wqe_shift);
+}
+
+static void mlx4_ib_srq_event(struct mlx4_srq *srq, enum mlx4_event type)
+{
+	struct ib_event event;
+	struct ib_srq *ibsrq = &to_mibsrq(srq)->ibsrq;
+
+	if (ibsrq->event_handler) {
+		event.device      = ibsrq->device;
+		event.element.srq = ibsrq;
+		switch (type) {
+		case MLX4_EVENT_TYPE_SRQ_LIMIT:
+			event.event = IB_EVENT_SRQ_LIMIT_REACHED;
+			break;
+		case MLX4_EVENT_TYPE_SRQ_CATAS_ERROR:
+			event.event = IB_EVENT_SRQ_ERR;
+			break;
+		default:
+			pr_warn("Unexpected event type %d "
+			       "on SRQ %06x\n", type, srq->srqn);
+			return;
+		}
+
+		ibsrq->event_handler(&event, ibsrq->srq_context);
+	}
+}
+
+struct ib_srq *mlx4_ib_create_srq(struct ib_pd *pd,
+				  struct ib_srq_init_attr *init_attr,
+				  struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_srq *srq;
+	struct mlx4_wqe_srq_next_seg *next;
+	struct mlx4_wqe_data_seg *scatter;
+	u32 cqn;
+	u16 xrcdn;
+	int desc_size;
+	int buf_size;
+	int err;
+	int i;
+
+	/* Sanity check SRQ size before proceeding */
+	if (init_attr->attr.max_wr  >= dev->dev->caps.max_srq_wqes ||
+	    init_attr->attr.max_sge >  dev->dev->caps.max_srq_sge)
+		return ERR_PTR(-EINVAL);
+
+	srq = kmalloc(sizeof *srq, GFP_KERNEL);
+	if (!srq)
+		return ERR_PTR(-ENOMEM);
+
+	mutex_init(&srq->mutex);
+	spin_lock_init(&srq->lock);
+	srq->msrq.max    = roundup_pow_of_two(init_attr->attr.max_wr + 1);
+	srq->msrq.max_gs = init_attr->attr.max_sge;
+
+	desc_size = max(32UL,
+			roundup_pow_of_two(sizeof (struct mlx4_wqe_srq_next_seg) +
+					   srq->msrq.max_gs *
+					   sizeof (struct mlx4_wqe_data_seg)));
+	srq->msrq.wqe_shift = ilog2(desc_size);
+
+	buf_size = srq->msrq.max * desc_size;
+
+	if (pd->uobject) {
+		struct mlx4_ib_create_srq ucmd;
+
+		if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd)) {
+			err = -EFAULT;
+			goto err_srq;
+		}
+
+		srq->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr,
+					buf_size, 0, 0);
+		if (IS_ERR(srq->umem)) {
+			err = PTR_ERR(srq->umem);
+			goto err_srq;
+		}
+
+		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(srq->umem),
+				    ilog2(srq->umem->page_size), &srq->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_ib_umem_write_mtt(dev, &srq->mtt, srq->umem);
+		if (err)
+			goto err_mtt;
+
+		err = mlx4_ib_db_map_user(to_mucontext(pd->uobject->context),
+					  ucmd.db_addr, &srq->db);
+		if (err)
+			goto err_mtt;
+	} else {
+		err = mlx4_db_alloc(dev->dev, &srq->db, 0);
+		if (err)
+			goto err_srq;
+
+		*srq->db.db = 0;
+
+		if (mlx4_buf_alloc(dev->dev, buf_size, PAGE_SIZE * 2, &srq->buf)) {
+			err = -ENOMEM;
+			goto err_db;
+		}
+
+		srq->head    = 0;
+		srq->tail    = srq->msrq.max - 1;
+		srq->wqe_ctr = 0;
+
+		for (i = 0; i < srq->msrq.max; ++i) {
+			next = get_wqe(srq, i);
+			next->next_wqe_index =
+				cpu_to_be16((i + 1) & (srq->msrq.max - 1));
+
+			for (scatter = (void *) (next + 1);
+			     (void *) scatter < (void *) next + desc_size;
+			     ++scatter)
+				scatter->lkey = cpu_to_be32(MLX4_INVALID_LKEY);
+		}
+
+		err = mlx4_mtt_init(dev->dev, srq->buf.npages, srq->buf.page_shift,
+				    &srq->mtt);
+		if (err)
+			goto err_buf;
+
+		err = mlx4_buf_write_mtt(dev->dev, &srq->mtt, &srq->buf);
+		if (err)
+			goto err_mtt;
+
+		srq->wrid = kmalloc(srq->msrq.max * sizeof (u64), GFP_KERNEL);
+		if (!srq->wrid) {
+			err = -ENOMEM;
+			goto err_mtt;
+		}
+	}
+
+	cqn = (init_attr->srq_type == IB_SRQT_XRC) ?
+		to_mcq(init_attr->ext.xrc.cq)->mcq.cqn : 0;
+	xrcdn = (init_attr->srq_type == IB_SRQT_XRC) ?
+		to_mxrcd(init_attr->ext.xrc.xrcd)->xrcdn :
+		(u16) dev->dev->caps.reserved_xrcds;
+	err = mlx4_srq_alloc(dev->dev, to_mpd(pd)->pdn, cqn, xrcdn, &srq->mtt,
+			     srq->db.dma, &srq->msrq);
+	if (err)
+		goto err_wrid;
+
+	srq->msrq.event = mlx4_ib_srq_event;
+	srq->ibsrq.ext.xrc.srq_num = srq->msrq.srqn;
+
+	if (pd->uobject)
+		if (ib_copy_to_udata(udata, &srq->msrq.srqn, sizeof (__u32))) {
+			err = -EFAULT;
+			goto err_wrid;
+		}
+
+	init_attr->attr.max_wr = srq->msrq.max - 1;
+
+	return &srq->ibsrq;
+
+err_wrid:
+	if (pd->uobject)
+		mlx4_ib_db_unmap_user(to_mucontext(pd->uobject->context), &srq->db);
+	else
+		kfree(srq->wrid);
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &srq->mtt);
+
+err_buf:
+	if (pd->uobject)
+		ib_umem_release(srq->umem);
+	else
+		mlx4_buf_free(dev->dev, buf_size, &srq->buf);
+
+err_db:
+	if (!pd->uobject)
+		mlx4_db_free(dev->dev, &srq->db);
+
+err_srq:
+	kfree(srq);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr,
+		       enum ib_srq_attr_mask attr_mask, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibsrq->device);
+	struct mlx4_ib_srq *srq = to_msrq(ibsrq);
+	int ret;
+
+	/* We don't support resizing SRQs (yet?) */
+	if (attr_mask & IB_SRQ_MAX_WR)
+		return -EINVAL;
+
+	if (attr_mask & IB_SRQ_LIMIT) {
+		if (attr->srq_limit >= srq->msrq.max)
+			return -EINVAL;
+
+		mutex_lock(&srq->mutex);
+		ret = mlx4_srq_arm(dev->dev, &srq->msrq, attr->srq_limit);
+		mutex_unlock(&srq->mutex);
+
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+int mlx4_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibsrq->device);
+	struct mlx4_ib_srq *srq = to_msrq(ibsrq);
+	int ret;
+	int limit_watermark;
+
+	ret = mlx4_srq_query(dev->dev, &srq->msrq, &limit_watermark);
+	if (ret)
+		return ret;
+
+	srq_attr->srq_limit = limit_watermark;
+	srq_attr->max_wr    = srq->msrq.max - 1;
+	srq_attr->max_sge   = srq->msrq.max_gs;
+
+	return 0;
+}
+
+int mlx4_ib_destroy_srq(struct ib_srq *srq)
+{
+	struct mlx4_ib_dev *dev = to_mdev(srq->device);
+	struct mlx4_ib_srq *msrq = to_msrq(srq);
+
+	mlx4_srq_free(dev->dev, &msrq->msrq);
+	mlx4_mtt_cleanup(dev->dev, &msrq->mtt);
+
+	if (srq->uobject) {
+		mlx4_ib_db_unmap_user(to_mucontext(srq->uobject->context), &msrq->db);
+		ib_umem_release(msrq->umem);
+	} else {
+		kfree(msrq->wrid);
+		mlx4_buf_free(dev->dev, msrq->msrq.max << msrq->msrq.wqe_shift,
+			      &msrq->buf);
+		mlx4_db_free(dev->dev, &msrq->db);
+	}
+
+	kfree(msrq);
+
+	return 0;
+}
+
+void mlx4_ib_free_srq_wqe(struct mlx4_ib_srq *srq, int wqe_index)
+{
+	struct mlx4_wqe_srq_next_seg *next;
+
+	/* always called with interrupts disabled. */
+	spin_lock(&srq->lock);
+
+	next = get_wqe(srq, srq->tail);
+	next->next_wqe_index = cpu_to_be16(wqe_index);
+	srq->tail = wqe_index;
+
+	spin_unlock(&srq->lock);
+}
+
+int mlx4_ib_post_srq_recv(struct ib_srq *ibsrq, struct ib_recv_wr *wr,
+			  struct ib_recv_wr **bad_wr)
+{
+	struct mlx4_ib_srq *srq = to_msrq(ibsrq);
+	struct mlx4_wqe_srq_next_seg *next;
+	struct mlx4_wqe_data_seg *scat;
+	unsigned long flags;
+	int err = 0;
+	int nreq;
+	int i;
+
+	spin_lock_irqsave(&srq->lock, flags);
+
+	for (nreq = 0; wr; ++nreq, wr = wr->next) {
+		if (unlikely(wr->num_sge > srq->msrq.max_gs)) {
+			err = -EINVAL;
+			*bad_wr = wr;
+			break;
+		}
+
+		if (unlikely(srq->head == srq->tail)) {
+			err = -ENOMEM;
+			*bad_wr = wr;
+			break;
+		}
+
+		srq->wrid[srq->head] = wr->wr_id;
+
+		next      = get_wqe(srq, srq->head);
+		srq->head = be16_to_cpu(next->next_wqe_index);
+		scat      = (struct mlx4_wqe_data_seg *) (next + 1);
+
+		for (i = 0; i < wr->num_sge; ++i) {
+			scat[i].byte_count = cpu_to_be32(wr->sg_list[i].length);
+			scat[i].lkey       = cpu_to_be32(wr->sg_list[i].lkey);
+			scat[i].addr       = cpu_to_be64(wr->sg_list[i].addr);
+		}
+
+		if (i < srq->msrq.max_gs) {
+			scat[i].byte_count = 0;
+			scat[i].lkey       = cpu_to_be32(MLX4_INVALID_LKEY);
+			scat[i].addr       = 0;
+		}
+	}
+
+	if (likely(nreq)) {
+		srq->wqe_ctr += nreq;
+
+		/*
+		 * Make sure that descriptors are written before
+		 * doorbell record.
+		 */
+		wmb();
+
+		*srq->db.db = cpu_to_be32(srq->wqe_ctr);
+	}
+
+	spin_unlock_irqrestore(&srq->lock, flags);
+
+	return err;
+}

drivers/infiniband/hw/mlx4/sysfs.c

@@ -0,0 +1,794 @@
+/*
+ * Copyright (c) 2012 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 "core_priv.h"*/
+#include "mlx4_ib.h"
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <linux/stat.h>
+
+#include <rdma/ib_mad.h>
+/*show_admin_alias_guid returns the administratively assigned value of that GUID.
+ * Values returned in buf parameter string:
+ *	0			- requests opensm to assign a value.
+ *	ffffffffffffffff	- delete this entry.
+ *	other			- value assigned by administrator.
+ */
+static ssize_t show_admin_alias_guid(struct device *dev,
+			      struct device_attribute *attr, char *buf)
+{
+	int record_num;/*0-15*/
+	int guid_index_in_rec; /*0 - 7*/
+	struct mlx4_ib_iov_sysfs_attr *mlx4_ib_iov_dentry =
+		container_of(attr, struct mlx4_ib_iov_sysfs_attr, dentry);
+	struct mlx4_ib_iov_port *port = mlx4_ib_iov_dentry->ctx;
+	struct mlx4_ib_dev *mdev = port->dev;
+
+	record_num = mlx4_ib_iov_dentry->entry_num / 8 ;
+	guid_index_in_rec = mlx4_ib_iov_dentry->entry_num % 8 ;
+
+	return sprintf(buf, "%llx\n",
+		       be64_to_cpu(*(__be64 *)&mdev->sriov.alias_guid.
+				   ports_guid[port->num - 1].
+				   all_rec_per_port[record_num].
+				   all_recs[8 * guid_index_in_rec]));
+}
+
+/* store_admin_alias_guid stores the (new) administratively assigned value of that GUID.
+ * Values in buf parameter string:
+ *	0			- requests opensm to assign a value.
+ *	0xffffffffffffffff	- delete this entry.
+ *	other			- guid value assigned by the administrator.
+ */
+static ssize_t store_admin_alias_guid(struct device *dev,
+				      struct device_attribute *attr,
+				      const char *buf, size_t count)
+{
+	int record_num;/*0-15*/
+	int guid_index_in_rec; /*0 - 7*/
+	struct mlx4_ib_iov_sysfs_attr *mlx4_ib_iov_dentry =
+		container_of(attr, struct mlx4_ib_iov_sysfs_attr, dentry);
+	struct mlx4_ib_iov_port *port = mlx4_ib_iov_dentry->ctx;
+	struct mlx4_ib_dev *mdev = port->dev;
+	u64 sysadmin_ag_val;
+
+	record_num = mlx4_ib_iov_dentry->entry_num / 8;
+	guid_index_in_rec = mlx4_ib_iov_dentry->entry_num % 8;
+	if (0 == record_num && 0 == guid_index_in_rec) {
+		pr_err("GUID 0 block 0 is RO\n");
+		return count;
+	}
+	sscanf(buf, "%llx", &sysadmin_ag_val);
+	*(__be64 *)&mdev->sriov.alias_guid.ports_guid[port->num - 1].
+		all_rec_per_port[record_num].
+		all_recs[GUID_REC_SIZE * guid_index_in_rec] =
+			cpu_to_be64(sysadmin_ag_val);
+
+	/* Change the state to be pending for update */
+	mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].status
+		= MLX4_GUID_INFO_STATUS_IDLE ;
+
+	mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].method
+		= MLX4_GUID_INFO_RECORD_SET;
+
+	switch (sysadmin_ag_val) {
+	case MLX4_GUID_FOR_DELETE_VAL:
+		mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].method
+			= MLX4_GUID_INFO_RECORD_DELETE;
+		mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].ownership
+			= MLX4_GUID_SYSADMIN_ASSIGN;
+		break;
+	/* The sysadmin requests the SM to re-assign */
+	case MLX4_NOT_SET_GUID:
+		mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].ownership
+			= MLX4_GUID_DRIVER_ASSIGN;
+		break;
+	/* The sysadmin requests a specific value.*/
+	default:
+		mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].ownership
+			= MLX4_GUID_SYSADMIN_ASSIGN;
+		break;
+	}
+
+	/* set the record index */
+	mdev->sriov.alias_guid.ports_guid[port->num - 1].all_rec_per_port[record_num].guid_indexes
+		= mlx4_ib_get_aguid_comp_mask_from_ix(guid_index_in_rec);
+
+	mlx4_ib_init_alias_guid_work(mdev, port->num - 1);
+
+	return count;
+}
+
+static ssize_t show_port_gid(struct device *dev,
+			     struct device_attribute *attr,
+			     char *buf)
+{
+	struct mlx4_ib_iov_sysfs_attr *mlx4_ib_iov_dentry =
+		container_of(attr, struct mlx4_ib_iov_sysfs_attr, dentry);
+	struct mlx4_ib_iov_port *port = mlx4_ib_iov_dentry->ctx;
+	struct mlx4_ib_dev *mdev = port->dev;
+	union ib_gid gid;
+	ssize_t ret;
+
+	ret = __mlx4_ib_query_gid(&mdev->ib_dev, port->num,
+				  mlx4_ib_iov_dentry->entry_num, &gid, 1);
+	if (ret)
+		return ret;
+	ret = sprintf(buf, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
+		      be16_to_cpu(((__be16 *) gid.raw)[0]),
+		      be16_to_cpu(((__be16 *) gid.raw)[1]),
+		      be16_to_cpu(((__be16 *) gid.raw)[2]),
+		      be16_to_cpu(((__be16 *) gid.raw)[3]),
+		      be16_to_cpu(((__be16 *) gid.raw)[4]),
+		      be16_to_cpu(((__be16 *) gid.raw)[5]),
+		      be16_to_cpu(((__be16 *) gid.raw)[6]),
+		      be16_to_cpu(((__be16 *) gid.raw)[7]));
+	return ret;
+}
+
+static ssize_t show_phys_port_pkey(struct device *dev,
+				   struct device_attribute *attr,
+				   char *buf)
+{
+	struct mlx4_ib_iov_sysfs_attr *mlx4_ib_iov_dentry =
+		container_of(attr, struct mlx4_ib_iov_sysfs_attr, dentry);
+	struct mlx4_ib_iov_port *port = mlx4_ib_iov_dentry->ctx;
+	struct mlx4_ib_dev *mdev = port->dev;
+	u16 pkey;
+	ssize_t ret;
+
+	ret = __mlx4_ib_query_pkey(&mdev->ib_dev, port->num,
+				   mlx4_ib_iov_dentry->entry_num, &pkey, 1);
+	if (ret)
+		return ret;
+
+	return sprintf(buf, "0x%04x\n", pkey);
+}
+
+#define DENTRY_REMOVE(_dentry)						\
+do {									\
+	sysfs_remove_file((_dentry)->kobj, &(_dentry)->dentry.attr);	\
+} while (0);
+
+static int create_sysfs_entry(void *_ctx, struct mlx4_ib_iov_sysfs_attr *_dentry,
+			      char *_name, struct kobject *_kobj,
+			      ssize_t (*show)(struct device *dev,
+					      struct device_attribute *attr,
+					      char *buf),
+			      ssize_t (*store)(struct device *dev,
+					       struct device_attribute *attr,
+					       const char *buf, size_t count)
+			      )
+{
+	int ret = 0;
+	struct mlx4_ib_iov_sysfs_attr *vdentry = _dentry;
+
+	vdentry->ctx = _ctx;
+	vdentry->dentry.show = show;
+	vdentry->dentry.store = store;
+	sysfs_attr_init(&vdentry->dentry.attr);
+	vdentry->dentry.attr.name = vdentry->name;
+	vdentry->dentry.attr.mode = 0;
+	vdentry->kobj = _kobj;
+	snprintf(vdentry->name, 15, "%s", _name);
+
+	if (vdentry->dentry.store)
+		vdentry->dentry.attr.mode |= S_IWUSR;
+
+	if (vdentry->dentry.show)
+		vdentry->dentry.attr.mode |= S_IRUGO;
+
+	ret = sysfs_create_file(vdentry->kobj, &vdentry->dentry.attr);
+	if (ret) {
+		pr_err("failed to create %s\n", vdentry->dentry.attr.name);
+		vdentry->ctx = NULL;
+		return ret;
+	}
+
+	return ret;
+}
+
+int add_sysfs_port_mcg_attr(struct mlx4_ib_dev *device, int port_num,
+		struct attribute *attr)
+{
+	struct mlx4_ib_iov_port *port = &device->iov_ports[port_num - 1];
+	int ret;
+
+	ret = sysfs_create_file(port->mcgs_parent, attr);
+	if (ret)
+		pr_err("failed to create %s\n", attr->name);
+
+	return ret;
+}
+
+void del_sysfs_port_mcg_attr(struct mlx4_ib_dev *device, int port_num,
+		struct attribute *attr)
+{
+	struct mlx4_ib_iov_port *port = &device->iov_ports[port_num - 1];
+
+	sysfs_remove_file(port->mcgs_parent, attr);
+}
+
+static int add_port_entries(struct mlx4_ib_dev *device, int port_num)
+{
+	int i;
+	char buff[10];
+	struct mlx4_ib_iov_port *port = NULL;
+	int ret = 0 ;
+	struct ib_port_attr attr;
+
+	/* get the physical gid and pkey table sizes.*/
+	ret = __mlx4_ib_query_port(&device->ib_dev, port_num, &attr, 1);
+	if (ret)
+		goto err;
+
+	port = &device->iov_ports[port_num - 1];
+	port->dev = device;
+	port->num = port_num;
+	/* Directory structure:
+	 * iov -
+	 *   port num -
+	 *	admin_guids
+	 *	gids (operational)
+	 *	mcg_table
+	 */
+	port->dentr_ar = kzalloc(sizeof (struct mlx4_ib_iov_sysfs_attr_ar),
+				 GFP_KERNEL);
+	if (!port->dentr_ar) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	sprintf(buff, "%d", port_num);
+	port->cur_port = kobject_create_and_add(buff,
+				 kobject_get(device->ports_parent));
+	if (!port->cur_port) {
+		ret = -ENOMEM;
+		goto kobj_create_err;
+	}
+	/* admin GUIDs */
+	port->admin_alias_parent = kobject_create_and_add("admin_guids",
+						  kobject_get(port->cur_port));
+	if (!port->admin_alias_parent) {
+		ret = -ENOMEM;
+		goto err_admin_guids;
+	}
+	for (i = 0 ; i < attr.gid_tbl_len; i++) {
+		sprintf(buff, "%d", i);
+		port->dentr_ar->dentries[i].entry_num = i;
+		ret = create_sysfs_entry(port, &port->dentr_ar->dentries[i],
+					  buff, port->admin_alias_parent,
+					  show_admin_alias_guid, store_admin_alias_guid);
+		if (ret)
+			goto err_admin_alias_parent;
+	}
+
+	/* gids subdirectory (operational gids) */
+	port->gids_parent = kobject_create_and_add("gids",
+						  kobject_get(port->cur_port));
+	if (!port->gids_parent) {
+		ret = -ENOMEM;
+		goto err_gids;
+	}
+
+	for (i = 0 ; i < attr.gid_tbl_len; i++) {
+		sprintf(buff, "%d", i);
+		port->dentr_ar->dentries[attr.gid_tbl_len + i].entry_num = i;
+		ret = create_sysfs_entry(port,
+					 &port->dentr_ar->dentries[attr.gid_tbl_len + i],
+					 buff,
+					 port->gids_parent, show_port_gid, NULL);
+		if (ret)
+			goto err_gids_parent;
+	}
+
+	/* physical port pkey table */
+	port->pkeys_parent =
+		kobject_create_and_add("pkeys", kobject_get(port->cur_port));
+	if (!port->pkeys_parent) {
+		ret = -ENOMEM;
+		goto err_pkeys;
+	}
+
+	for (i = 0 ; i < attr.pkey_tbl_len; i++) {
+		sprintf(buff, "%d", i);
+		port->dentr_ar->dentries[2 * attr.gid_tbl_len + i].entry_num = i;
+		ret = create_sysfs_entry(port,
+					 &port->dentr_ar->dentries[2 * attr.gid_tbl_len + i],
+					 buff, port->pkeys_parent,
+					 show_phys_port_pkey, NULL);
+		if (ret)
+			goto err_pkeys_parent;
+	}
+
+	/* MCGs table */
+	port->mcgs_parent =
+		kobject_create_and_add("mcgs", kobject_get(port->cur_port));
+	if (!port->mcgs_parent) {
+		ret = -ENOMEM;
+		goto err_mcgs;
+	}
+	return 0;
+
+err_mcgs:
+	kobject_put(port->cur_port);
+
+err_pkeys_parent:
+	kobject_put(port->pkeys_parent);
+
+err_pkeys:
+	kobject_put(port->cur_port);
+
+err_gids_parent:
+	kobject_put(port->gids_parent);
+
+err_gids:
+	kobject_put(port->cur_port);
+
+err_admin_alias_parent:
+	kobject_put(port->admin_alias_parent);
+
+err_admin_guids:
+	kobject_put(port->cur_port);
+	kobject_put(port->cur_port); /* once more for create_and_add buff */
+
+kobj_create_err:
+	kobject_put(device->ports_parent);
+	kfree(port->dentr_ar);
+
+err:
+	pr_err("add_port_entries FAILED: for port:%d, error: %d\n",
+	       port_num, ret);
+	return ret;
+}
+
+static void get_name(struct mlx4_ib_dev *dev, char *name, int i, int max)
+{
+	char base_name[9];
+
+	/* pci_name format is: bus:dev:func -> xxxx:yy:zz.n */
+	strlcpy(name, pci_name(dev->dev->pdev), max);
+	strncpy(base_name, name, 8); /*till xxxx:yy:*/
+	base_name[8] = '\0';
+	/* with no ARI only 3 last bits are used so when the fn is higher than 8
+	 * need to add it to the dev num, so count in the last number will be
+	 * modulo 8 */
+	sprintf(name, "%s%.2d.%d", base_name, (i/8), (i%8));
+}
+
+struct mlx4_port {
+	struct kobject         kobj;
+	struct mlx4_ib_dev    *dev;
+	struct attribute_group pkey_group;
+	struct attribute_group gid_group;
+	u8                     port_num;
+	int		       slave;
+};
+
+
+static void mlx4_port_release(struct kobject *kobj)
+{
+	struct mlx4_port *p = container_of(kobj, struct mlx4_port, kobj);
+	struct attribute *a;
+	int i;
+
+	for (i = 0; (a = p->pkey_group.attrs[i]); ++i)
+		kfree(a);
+	kfree(p->pkey_group.attrs);
+	for (i = 0; (a = p->gid_group.attrs[i]); ++i)
+		kfree(a);
+	kfree(p->gid_group.attrs);
+	kfree(p);
+}
+
+struct port_attribute {
+	struct attribute attr;
+	ssize_t (*show)(struct mlx4_port *, struct port_attribute *, char *buf);
+	ssize_t (*store)(struct mlx4_port *, struct port_attribute *,
+			 const char *buf, size_t count);
+};
+
+static ssize_t port_attr_show(struct kobject *kobj,
+			      struct attribute *attr, char *buf)
+{
+	struct port_attribute *port_attr =
+		container_of(attr, struct port_attribute, attr);
+	struct mlx4_port *p = container_of(kobj, struct mlx4_port, kobj);
+
+	if (!port_attr->show)
+		return -EIO;
+	return port_attr->show(p, port_attr, buf);
+}
+
+static ssize_t port_attr_store(struct kobject *kobj,
+			       struct attribute *attr,
+			       const char *buf, size_t size)
+{
+	struct port_attribute *port_attr =
+		container_of(attr, struct port_attribute, attr);
+	struct mlx4_port *p = container_of(kobj, struct mlx4_port, kobj);
+
+	if (!port_attr->store)
+		return -EIO;
+	return port_attr->store(p, port_attr, buf, size);
+}
+
+static const struct sysfs_ops port_sysfs_ops = {
+	.show = port_attr_show,
+	.store = port_attr_store,
+};
+
+static struct kobj_type port_type = {
+	.release    = mlx4_port_release,
+	.sysfs_ops  = &port_sysfs_ops,
+};
+
+struct port_table_attribute {
+	struct port_attribute	attr;
+	char			name[8];
+	int			index;
+};
+
+static ssize_t show_port_pkey(struct mlx4_port *p, struct port_attribute *attr,
+			      char *buf)
+{
+	struct port_table_attribute *tab_attr =
+		container_of(attr, struct port_table_attribute, attr);
+	ssize_t ret = -ENODEV;
+
+	if (p->dev->pkeys.virt2phys_pkey[p->slave][p->port_num - 1][tab_attr->index] >=
+	    (p->dev->dev->caps.pkey_table_len[p->port_num]))
+		ret = sprintf(buf, "none\n");
+	else
+		ret = sprintf(buf, "%d\n",
+			      p->dev->pkeys.virt2phys_pkey[p->slave]
+			      [p->port_num - 1][tab_attr->index]);
+	return ret;
+}
+
+static ssize_t store_port_pkey(struct mlx4_port *p, struct port_attribute *attr,
+			       const char *buf, size_t count)
+{
+	struct port_table_attribute *tab_attr =
+		container_of(attr, struct port_table_attribute, attr);
+	int idx;
+	int err;
+
+	/* do not allow remapping Dom0 virtual pkey table */
+	if (p->slave == mlx4_master_func_num(p->dev->dev))
+		return -EINVAL;
+
+	if (!strncasecmp(buf, "no", 2))
+		idx = p->dev->dev->phys_caps.pkey_phys_table_len[p->port_num] - 1;
+	else if (sscanf(buf, "%i", &idx) != 1 ||
+		 idx >= p->dev->dev->caps.pkey_table_len[p->port_num] ||
+		 idx < 0)
+		return -EINVAL;
+
+	p->dev->pkeys.virt2phys_pkey[p->slave][p->port_num - 1]
+				    [tab_attr->index] = idx;
+	mlx4_sync_pkey_table(p->dev->dev, p->slave, p->port_num,
+			     tab_attr->index, idx);
+	err = mlx4_gen_pkey_eqe(p->dev->dev, p->slave, p->port_num);
+	if (err) {
+		pr_err("mlx4_gen_pkey_eqe failed for slave %d,"
+		       " port %d, index %d\n", p->slave, p->port_num, idx);
+		return err;
+	}
+	return count;
+}
+
+static ssize_t show_port_gid_idx(struct mlx4_port *p,
+				 struct port_attribute *attr, char *buf)
+{
+	return sprintf(buf, "%d\n", p->slave);
+}
+
+static struct attribute **
+alloc_group_attrs(ssize_t (*show)(struct mlx4_port *,
+				  struct port_attribute *, char *buf),
+		  ssize_t (*store)(struct mlx4_port *, struct port_attribute *,
+				   const char *buf, size_t count),
+		  int len)
+{
+	struct attribute **tab_attr;
+	struct port_table_attribute *element;
+	int i;
+
+	tab_attr = kcalloc(1 + len, sizeof (struct attribute *), GFP_KERNEL);
+	if (!tab_attr)
+		return NULL;
+
+	for (i = 0; i < len; i++) {
+		element = kzalloc(sizeof (struct port_table_attribute),
+				  GFP_KERNEL);
+		if (!element)
+			goto err;
+		if (snprintf(element->name, sizeof (element->name),
+			     "%d", i) >= sizeof (element->name)) {
+			kfree(element);
+			goto err;
+		}
+		sysfs_attr_init(&element->attr.attr);
+		element->attr.attr.name  = element->name;
+		if (store) {
+			element->attr.attr.mode  = S_IWUSR | S_IRUGO;
+			element->attr.store	 = store;
+		} else
+			element->attr.attr.mode  = S_IRUGO;
+
+		element->attr.show       = show;
+		element->index		 = i;
+		tab_attr[i] = &element->attr.attr;
+	}
+	return tab_attr;
+
+err:
+	while (--i >= 0)
+		kfree(tab_attr[i]);
+	kfree(tab_attr);
+	return NULL;
+}
+
+static int add_port(struct mlx4_ib_dev *dev, int port_num, int slave)
+{
+	struct mlx4_port *p;
+	int i;
+	int ret;
+
+	p = kzalloc(sizeof *p, GFP_KERNEL);
+	if (!p)
+		return -ENOMEM;
+
+	p->dev = dev;
+	p->port_num = port_num;
+	p->slave = slave;
+
+	ret = kobject_init_and_add(&p->kobj, &port_type,
+				   kobject_get(dev->dev_ports_parent[slave]),
+				   "%d", port_num);
+	if (ret)
+		goto err_alloc;
+
+	p->pkey_group.name  = "pkey_idx";
+	p->pkey_group.attrs =
+		alloc_group_attrs(show_port_pkey, store_port_pkey,
+				  dev->dev->caps.pkey_table_len[port_num]);
+	if (!p->pkey_group.attrs)
+		goto err_alloc;
+
+	ret = sysfs_create_group(&p->kobj, &p->pkey_group);
+	if (ret)
+		goto err_free_pkey;
+
+	p->gid_group.name  = "gid_idx";
+	p->gid_group.attrs = alloc_group_attrs(show_port_gid_idx, NULL, 1);
+	if (!p->gid_group.attrs)
+		goto err_free_pkey;
+
+	ret = sysfs_create_group(&p->kobj, &p->gid_group);
+	if (ret)
+		goto err_free_gid;
+
+	list_add_tail(&p->kobj.entry, &dev->pkeys.pkey_port_list[slave]);
+	return 0;
+
+err_free_gid:
+	kfree(p->gid_group.attrs[0]);
+	kfree(p->gid_group.attrs);
+
+err_free_pkey:
+	for (i = 0; i < dev->dev->caps.pkey_table_len[port_num]; ++i)
+		kfree(p->pkey_group.attrs[i]);
+	kfree(p->pkey_group.attrs);
+
+err_alloc:
+	kobject_put(dev->dev_ports_parent[slave]);
+	kfree(p);
+	return ret;
+}
+
+static int register_one_pkey_tree(struct mlx4_ib_dev *dev, int slave)
+{
+	char name[32];
+	int err;
+	int port;
+	struct kobject *p, *t;
+	struct mlx4_port *mport;
+
+	get_name(dev, name, slave, sizeof name);
+
+	dev->pkeys.device_parent[slave] =
+		kobject_create_and_add(name, kobject_get(dev->iov_parent));
+
+	if (!dev->pkeys.device_parent[slave]) {
+		err = -ENOMEM;
+		goto fail_dev;
+	}
+
+	INIT_LIST_HEAD(&dev->pkeys.pkey_port_list[slave]);
+
+	dev->dev_ports_parent[slave] =
+		kobject_create_and_add("ports",
+				       kobject_get(dev->pkeys.device_parent[slave]));
+
+	if (!dev->dev_ports_parent[slave]) {
+		err = -ENOMEM;
+		goto err_ports;
+	}
+
+	for (port = 1; port <= dev->dev->caps.num_ports; ++port) {
+		err = add_port(dev, port, slave);
+		if (err)
+			goto err_add;
+	}
+	return 0;
+
+err_add:
+	list_for_each_entry_safe(p, t,
+				 &dev->pkeys.pkey_port_list[slave],
+				 entry) {
+		list_del(&p->entry);
+		mport = container_of(p, struct mlx4_port, kobj);
+		sysfs_remove_group(p, &mport->pkey_group);
+		sysfs_remove_group(p, &mport->gid_group);
+		kobject_put(p);
+	}
+	kobject_put(dev->dev_ports_parent[slave]);
+
+err_ports:
+	kobject_put(dev->pkeys.device_parent[slave]);
+	/* extra put for the device_parent create_and_add */
+	kobject_put(dev->pkeys.device_parent[slave]);
+
+fail_dev:
+	kobject_put(dev->iov_parent);
+	return err;
+}
+
+static int register_pkey_tree(struct mlx4_ib_dev *device)
+{
+	int i;
+
+	if (!mlx4_is_master(device->dev))
+		return 0;
+
+	for (i = 0; i <= device->dev->num_vfs; ++i)
+		register_one_pkey_tree(device, i);
+
+	return 0;
+}
+
+static void unregister_pkey_tree(struct mlx4_ib_dev *device)
+{
+	int slave;
+	struct kobject *p, *t;
+	struct mlx4_port *port;
+
+	if (!mlx4_is_master(device->dev))
+		return;
+
+	for (slave = device->dev->num_vfs; slave >= 0; --slave) {
+		list_for_each_entry_safe(p, t,
+					 &device->pkeys.pkey_port_list[slave],
+					 entry) {
+			list_del(&p->entry);
+			port = container_of(p, struct mlx4_port, kobj);
+			sysfs_remove_group(p, &port->pkey_group);
+			sysfs_remove_group(p, &port->gid_group);
+			kobject_put(p);
+			kobject_put(device->dev_ports_parent[slave]);
+		}
+		kobject_put(device->dev_ports_parent[slave]);
+		kobject_put(device->pkeys.device_parent[slave]);
+		kobject_put(device->pkeys.device_parent[slave]);
+		kobject_put(device->iov_parent);
+	}
+}
+
+int mlx4_ib_device_register_sysfs(struct mlx4_ib_dev *dev)
+{
+	int i;
+	int ret = 0;
+
+	if (!mlx4_is_master(dev->dev))
+		return 0;
+
+	dev->iov_parent =
+		kobject_create_and_add("iov",
+				       kobject_get(dev->ib_dev.ports_parent->parent));
+	if (!dev->iov_parent) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	dev->ports_parent =
+		kobject_create_and_add("ports",
+				       kobject_get(dev->iov_parent));
+	if (!dev->ports_parent) {
+		ret = -ENOMEM;
+		goto err_ports;
+	}
+
+	for (i = 1; i <= dev->ib_dev.phys_port_cnt; ++i) {
+		ret = add_port_entries(dev, i);
+		if (ret)
+			goto err_add_entries;
+	}
+
+	ret = register_pkey_tree(dev);
+	if (ret)
+		goto err_add_entries;
+	return 0;
+
+err_add_entries:
+	kobject_put(dev->ports_parent);
+
+err_ports:
+	kobject_put(dev->iov_parent);
+err:
+	kobject_put(dev->ib_dev.ports_parent->parent);
+	pr_err("mlx4_ib_device_register_sysfs error (%d)\n", ret);
+	return ret;
+}
+
+static void unregister_alias_guid_tree(struct mlx4_ib_dev *device)
+{
+	struct mlx4_ib_iov_port *p;
+	int i;
+
+	if (!mlx4_is_master(device->dev))
+		return;
+
+	for (i = 0; i < device->dev->caps.num_ports; i++) {
+		p = &device->iov_ports[i];
+		kobject_put(p->admin_alias_parent);
+		kobject_put(p->gids_parent);
+		kobject_put(p->pkeys_parent);
+		kobject_put(p->mcgs_parent);
+		kobject_put(p->cur_port);
+		kobject_put(p->cur_port);
+		kobject_put(p->cur_port);
+		kobject_put(p->cur_port);
+		kobject_put(p->cur_port);
+		kobject_put(p->dev->ports_parent);
+		kfree(p->dentr_ar);
+	}
+}
+
+void mlx4_ib_device_unregister_sysfs(struct mlx4_ib_dev *device)
+{
+	unregister_alias_guid_tree(device);
+	unregister_pkey_tree(device);
+	kobject_put(device->ports_parent);
+	kobject_put(device->iov_parent);
+	kobject_put(device->iov_parent);
+	kobject_put(device->ib_dev.ports_parent->parent);
+}

drivers/infiniband/hw/mlx4/user.h

@@ -0,0 +1,107 @@
+/*
+ * Copyright (c) 2007 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2007, 2008 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.
+ */
+
+#ifndef MLX4_IB_USER_H
+#define MLX4_IB_USER_H
+
+#include <linux/types.h>
+
+/*
+ * Increment this value if any changes that break userspace ABI
+ * compatibility are made.
+ */
+
+#define MLX4_IB_UVERBS_NO_DEV_CAPS_ABI_VERSION	3
+#define MLX4_IB_UVERBS_ABI_VERSION		4
+
+/*
+ * Make sure that all structs defined in this file remain laid out so
+ * that they pack the same way on 32-bit and 64-bit architectures (to
+ * avoid incompatibility between 32-bit userspace and 64-bit kernels).
+ * In particular do not use pointer types -- pass pointers in __u64
+ * instead.
+ */
+
+struct mlx4_ib_alloc_ucontext_resp_v3 {
+	__u32	qp_tab_size;
+	__u16	bf_reg_size;
+	__u16	bf_regs_per_page;
+};
+
+struct mlx4_ib_alloc_ucontext_resp {
+	__u32	dev_caps;
+	__u32	qp_tab_size;
+	__u16	bf_reg_size;
+	__u16	bf_regs_per_page;
+	__u32	cqe_size;
+};
+
+struct mlx4_ib_alloc_pd_resp {
+	__u32	pdn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_create_cq {
+	__u64	buf_addr;
+	__u64	db_addr;
+};
+
+struct mlx4_ib_create_cq_resp {
+	__u32	cqn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_resize_cq {
+	__u64	buf_addr;
+};
+
+struct mlx4_ib_create_srq {
+	__u64	buf_addr;
+	__u64	db_addr;
+};
+
+struct mlx4_ib_create_srq_resp {
+	__u32	srqn;
+	__u32	reserved;
+};
+
+struct mlx4_ib_create_qp {
+	__u64	buf_addr;
+	__u64	db_addr;
+	__u8	log_sq_bb_count;
+	__u8	log_sq_stride;
+	__u8	sq_no_prefetch;
+	__u8	reserved[5];
+};
+
+#endif /* MLX4_IB_USER_H */