[dpdk-dev] [PATCH 5/5] net/mlx4: add new Memory Region support

[Yongseok Koh <yskoh@mellanox.com>]

This is the new design of Memory Region (MR) for mlx PMD, in order to:
- Accommodate the new memory hotplug model.
- Support non-contiguous Mempool.

Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
---
 drivers/net/mlx4/mlx4.c      |   26 +
 drivers/net/mlx4/mlx4.h      |   14 +
 drivers/net/mlx4/mlx4_mr.c   | 1165 ++++++++++++++++++++++++++++++++++++++++++
 drivers/net/mlx4/mlx4_mr.h   |  126 +++++
 drivers/net/mlx4/mlx4_rxq.c  |   14 +-
 drivers/net/mlx4/mlx4_rxtx.c |   35 +-
 drivers/net/mlx4/mlx4_rxtx.h |   71 ++-
 drivers/net/mlx4/mlx4_txq.c  |    8 +
 8 files changed, 1428 insertions(+), 31 deletions(-)
 create mode 100644 drivers/net/mlx4/mlx4_mr.h

diff --git a/drivers/net/mlx4/mlx4.c b/drivers/net/mlx4/mlx4.c
index 47451b651..e6c701d13 100644
--- a/drivers/net/mlx4/mlx4.c
+++ b/drivers/net/mlx4/mlx4.c
@@ -44,9 +44,15 @@
 #include "mlx4.h"
 #include "mlx4_glue.h"
 #include "mlx4_flow.h"
+#include "mlx4_mr.h"
 #include "mlx4_rxtx.h"
 #include "mlx4_utils.h"
 
+struct mlx4_dev_list mlx4_mem_event_cb_list =
+	LIST_HEAD_INITIALIZER(mlx4_mem_event_cb_list);
+
+rte_rwlock_t mlx4_mem_event_rwlock = RTE_RWLOCK_INITIALIZER;
+
 /** Configuration structure for device arguments. */
 struct mlx4_conf {
 	struct {
@@ -92,6 +98,20 @@ mlx4_dev_configure(struct rte_eth_dev *dev)
 	if (ret)
 		ERROR("%p: interrupt handler installation failed",
 		      (void *)dev);
+	/*
+	 * Once the device is added to the list of memory event callback, its
+	 * global MR cache table cannot be expanded on the fly because of
+	 * deadlock. If it overflows, lookup should be done by searching MR list
+	 * linearly, which is slow.
+	 */
+	if (mlx4_mr_btree_init(&priv->mr.cache, MLX4_MR_BTREE_CACHE_N * 2,
+			       dev->device->numa_node)) {
+		/* rte_errno is already set. */
+		return -rte_errno;
+	}
+	rte_rwlock_write_lock(&mlx4_mem_event_rwlock);
+	LIST_INSERT_HEAD(&mlx4_mem_event_cb_list, priv, mem_event_cb);
+	rte_rwlock_write_unlock(&mlx4_mem_event_rwlock);
 exit:
 	return ret;
 }
@@ -125,6 +145,9 @@ mlx4_dev_start(struct rte_eth_dev *dev)
 		      (void *)dev, strerror(-ret));
 		goto err;
 	}
+#ifndef NDEBUG
+	mlx4_mr_dump_dev(dev);
+#endif
 	ret = mlx4_rxq_intr_enable(priv);
 	if (ret) {
 		ERROR("%p: interrupt handler installation failed",
@@ -200,6 +223,7 @@ mlx4_dev_close(struct rte_eth_dev *dev)
 		mlx4_rx_queue_release(dev->data->rx_queues[i]);
 	for (i = 0; i != dev->data->nb_tx_queues; ++i)
 		mlx4_tx_queue_release(dev->data->tx_queues[i]);
+	mlx4_mr_release(dev);
 	if (priv->pd != NULL) {
 		assert(priv->ctx != NULL);
 		claim_zero(mlx4_glue->dealloc_pd(priv->pd));
@@ -964,6 +988,8 @@ rte_mlx4_pmd_init(void)
 	}
 	mlx4_glue->fork_init();
 	rte_pci_register(&mlx4_driver);
+	rte_mem_event_callback_register("MLX4_MEM_EVENT_CB",
+					mlx4_mr_mem_event_cb);
 }
 
 RTE_PMD_EXPORT_NAME(net_mlx4, __COUNTER__);
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
index e0e1b5d4c..300cb4d7a 100644
--- a/drivers/net/mlx4/mlx4.h
+++ b/drivers/net/mlx4/mlx4.h
@@ -23,6 +23,9 @@
 #include <rte_ether.h>
 #include <rte_interrupts.h>
 #include <rte_mempool.h>
+#include <rte_rwlock.h>
+
+#include "mlx4_mr.h"
 
 #ifndef IBV_RX_HASH_INNER
 /** This is not necessarily defined by supported RDMA core versions. */
@@ -66,8 +69,12 @@ struct rxq;
 struct txq;
 struct rte_flow;
 
+LIST_HEAD(mlx4_dev_list, priv);
+LIST_HEAD(mlx4_mr_list, mlx4_mr);
+
 /** Private data structure. */
 struct priv {
+	LIST_ENTRY(priv) mem_event_cb; /* Called by memory event callback. */
 	struct rte_eth_dev *dev; /**< Ethernet device. */
 	struct ibv_context *ctx; /**< Verbs context. */
 	struct ibv_device_attr device_attr; /**< Device properties. */
@@ -86,6 +93,13 @@ struct priv {
 	uint64_t hw_rss_sup; /**< Supported RSS hash fields (Verbs format). */
 	struct rte_intr_handle intr_handle; /**< Port interrupt handle. */
 	struct mlx4_drop *drop; /**< Shared resources for drop flow rules. */
+	struct {
+		uint32_t dev_gen; /* Generation number to flush local caches. */
+		rte_rwlock_t rwlock; /* MR Lock. */
+		struct mlx4_mr_btree cache; /* Global MR cache table. */
+		struct mlx4_mr_list mr_list; /* Registered MR list. */
+		struct mlx4_mr_list mr_free_list; /* Freed MR list. */
+	} mr;
 	LIST_HEAD(, mlx4_rss) rss; /**< Shared targets for Rx flow rules. */
 	LIST_HEAD(, rte_flow) flows; /**< Configured flow rule handles. */
 	struct ether_addr mac[MLX4_MAX_MAC_ADDRESSES];
diff --git a/drivers/net/mlx4/mlx4_mr.c b/drivers/net/mlx4/mlx4_mr.c
index 3c87f6849..4812f7109 100644
--- a/drivers/net/mlx4/mlx4_mr.c
+++ b/drivers/net/mlx4/mlx4_mr.c
@@ -30,8 +30,1173 @@
 #include <rte_malloc.h>
 #include <rte_memory.h>
 #include <rte_mempool.h>
+#include <rte_rwlock.h>
 
 #include "mlx4_glue.h"
+#include "mlx4_mr.h"
 #include "mlx4_rxtx.h"
 #include "mlx4_utils.h"
 
+struct mr_find_contig_memsegs_data {
+	uintptr_t addr;
+	uintptr_t start;
+	uintptr_t end;
+	const struct rte_memseg_list *msl;
+};
+
+struct mr_update_mp_data {
+	struct rte_eth_dev *dev;
+	struct mlx4_mr_ctrl *mr_ctrl;
+	int ret;
+};
+
+/**
+ * Expand B-tree table to a given size. Can't be called with holding
+ * memory_hotplug_lock or priv->mr.rwlock due to rte_realloc().
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ * @param n
+ *   Number of entries for expansion.
+ *
+ * @return
+ *   0 on success, -1 on failure.
+ */
+static int
+mr_btree_expand(struct mlx4_mr_btree *bt, int n)
+{
+	void *mem;
+	int ret = 0;
+
+	if (n <= bt->size)
+		return ret;
+	/*
+	 * Downside of directly using rte_realloc() is that SOCKET_ID_ANY is
+	 * used inside if there's no room to expand. Because this is a quite
+	 * rare case and a part of very slow path, it is very acceptable.
+	 * Initially cache_bh[] will be given practically enough space and once
+	 * it is expanded, expansion wouldn't be needed again ever.
+	 */
+	mem = rte_realloc(bt->table, n * sizeof(struct mlx4_mr_cache), 0);
+	if (mem == NULL) {
+		/* Not an error, B-tree search will be skipped. */
+		WARN("failed to expand MR B-tree (%p) table", (void *)bt);
+		ret = -1;
+	} else {
+		DEBUG("expanded MR B-tree table (size=%u)", n);
+		bt->table = mem;
+		bt->size = n;
+	}
+	return ret;
+}
+
+/**
+ * Look up LKey from given B-tree lookup table, store the last index and return
+ * searched LKey.
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ * @param[out] idx
+ *   Pointer to index. Even on searh failure, returns index where it stops
+ *   searching so that index can be used when inserting a new entry.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static uint32_t
+mr_btree_lookup(struct mlx4_mr_btree *bt, uint16_t *idx, uintptr_t addr)
+{
+	struct mlx4_mr_cache *lkp_tbl;
+	uint16_t n;
+	uint16_t base = 0;
+
+	assert(bt != NULL);
+	lkp_tbl = *bt->table;
+	n = bt->len;
+	/* First entry must be NULL for comparison. */
+	assert(bt->len > 0 || (lkp_tbl[0].start == 0 &&
+			       lkp_tbl[0].lkey == UINT32_MAX));
+	/* Binary search. */
+	do {
+		register uint16_t delta = n >> 1;
+
+		if (addr < lkp_tbl[base + delta].start) {
+			n = delta;
+		} else {
+			base += delta;
+			n -= delta;
+		}
+	} while (n > 1);
+	assert(addr >= lkp_tbl[base].start);
+	*idx = base;
+	if (addr < lkp_tbl[base].end)
+		return lkp_tbl[base].lkey;
+	/* Not found. */
+	return UINT32_MAX;
+}
+
+/**
+ * Insert an entry to B-tree lookup table.
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ * @param entry
+ *   Pointer to new entry to insert.
+ *
+ * @return
+ *   0 on success, -1 on failure.
+ */
+static int
+mr_btree_insert(struct mlx4_mr_btree *bt, struct mlx4_mr_cache *entry)
+{
+	struct mlx4_mr_cache *lkp_tbl;
+	uint16_t idx = 0;
+	size_t shift;
+
+	assert(bt != NULL);
+	assert(bt->len <= bt->size);
+	assert(bt->len > 0);
+	lkp_tbl = *bt->table;
+	/* Find out the slot for insertion. */
+	if (mr_btree_lookup(bt, &idx, entry->start) != UINT32_MAX) {
+		DEBUG("abort insertion to B-tree(%p):"
+		      " already exist at idx=%u [0x%lx, 0x%lx) lkey=0x%x",
+		      (void *)bt, idx, entry->start, entry->end, entry->lkey);
+		/* Already exist, return. */
+		return 0;
+	}
+	/* If table is full, return error. */
+	if (unlikely(bt->len == bt->size)) {
+		bt->overflow = 1;
+		return -1;
+	}
+	/* Insert entry. */
+	++idx;
+	shift = (bt->len - idx) * sizeof(struct mlx4_mr_cache);
+	if (shift)
+		memmove(&lkp_tbl[idx + 1], &lkp_tbl[idx], shift);
+	lkp_tbl[idx] = *entry;
+	bt->len++;
+	DEBUG("inserted B-tree(%p)[%u], [0x%lx, 0x%lx) lkey=0x%x",
+	      (void *)bt, idx, entry->start, entry->end, entry->lkey);
+	return 0;
+}
+
+/**
+ * Initialize B-tree and allocate memory for lookup table.
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ * @param n
+ *   Number of entries to allocate.
+ * @param socket
+ *   NUMA socket on which memory must be allocated.
+ *
+ * @return
+ *   0 on success, a negative errno value otherwise and rte_errno is set.
+ */
+int
+mlx4_mr_btree_init(struct mlx4_mr_btree *bt, int n, int socket)
+{
+	if (bt == NULL) {
+		rte_errno = EINVAL;
+		return -rte_errno;
+	}
+	memset(bt, 0, sizeof(*bt));
+	bt->table = rte_calloc_socket("B-tree table",
+				      n, sizeof(struct mlx4_mr_cache),
+				      0, socket);
+	if (bt->table == NULL) {
+		rte_errno = ENOMEM;
+		ERROR("failed to allocate memory for btree cache on socket %d",
+		      socket);
+		return -rte_errno;
+	}
+	bt->size = n;
+	/* First entry must be NULL for binary search. */
+	(*bt->table)[bt->len++] = (struct mlx4_mr_cache) {
+		.lkey = UINT32_MAX,
+	};
+	DEBUG("initialized B-tree %p with table %p",
+	      (void *)bt, (void *)bt->table);
+	return 0;
+}
+
+/**
+ * Free B-tree resources.
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ */
+void
+mlx4_mr_btree_free(struct mlx4_mr_btree *bt)
+{
+	if (bt == NULL)
+		return;
+	DEBUG("freeing B-tree %p with table %p", (void *)bt, (void *)bt->table);
+	rte_free(bt->table);
+	memset(bt, 0, sizeof(*bt));
+}
+
+#ifndef NDEBUG
+/**
+ * Dump all the entries in a B-tree
+ *
+ * @param bt
+ *   Pointer to B-tree structure.
+ */
+void
+mlx4_mr_btree_dump(struct mlx4_mr_btree *bt)
+{
+	int idx;
+	struct mlx4_mr_cache *lkp_tbl;
+
+	if (bt == NULL)
+		return;
+	lkp_tbl = *bt->table;
+	for (idx = 0; idx < bt->len; ++idx) {
+		struct mlx4_mr_cache *entry = &lkp_tbl[idx];
+
+		DEBUG("B-tree(%p)[%u], [0x%lx, 0x%lx) lkey=0x%x",
+		      (void *)bt, idx, entry->start, entry->end, entry->lkey);
+	}
+}
+#endif
+
+/**
+ * Find virtually contiguous memory chunk in a given MR.
+ *
+ * @param dev
+ *   Pointer to MR structure.
+ * @param[out] entry
+ *   Pointer to returning MR cache entry. If not found, this will not be
+ *   updated.
+ * @param start_idx
+ *   Start index of the memseg bitmap.
+ *
+ * @return
+ *   Next index to go on lookup.
+ */
+static int
+mr_find_next_chunk(struct mlx4_mr *mr, struct mlx4_mr_cache *entry,
+		   int base_idx)
+{
+	uintptr_t start = 0;
+	uintptr_t end = 0;
+	uint32_t idx = 0;
+
+	for (idx = base_idx; idx < mr->ms_bmp_n; ++idx) {
+		if (rte_bitmap_get(mr->ms_bmp, idx)) {
+			const struct rte_memseg_list *msl;
+			const struct rte_memseg *ms;
+
+			msl = mr->msl;
+			ms = rte_fbarray_get(&msl->memseg_arr,
+					     mr->ms_base_idx + idx);
+			assert(msl->page_sz == ms->hugepage_sz);
+			if (!start)
+				start = ms->addr_64;
+			end = ms->addr_64 + ms->hugepage_sz;
+		} else if (start) {
+			/* Passed the end of a fragment. */
+			break;
+		}
+	}
+	if (start) {
+		/* Found one chunk. */
+		entry->start = start;
+		entry->end = end;
+		entry->lkey = rte_cpu_to_be_32(mr->ibv_mr->lkey);
+	}
+	return idx;
+}
+
+/**
+ * Insert a MR to the global B-tree cache. It may fail due to low-on-memory.
+ * Then, this entry will have to be searched by mr_lookup_dev_list() in
+ * mlx4_mr_create() on miss.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param mr
+ *   Pointer to MR to insert.
+ *
+ * @return
+ *   0 on success, -1 on failure.
+ */
+static int
+mr_insert_dev_cache(struct rte_eth_dev *dev, struct mlx4_mr *mr)
+{
+	struct priv *priv = dev->data->dev_private;
+	unsigned int n;
+
+	DEBUG("port %u inserting MR(%p) to global cache",
+	      dev->data->port_id, (void *)mr);
+	for (n = 0; n < mr->ms_bmp_n; ) {
+		struct mlx4_mr_cache entry = { 0, };
+
+		/* Find a contiguous chunk and advance the index. */
+		n = mr_find_next_chunk(mr, &entry, n);
+		if (!entry.end)
+			break;
+		if (mr_btree_insert(&priv->mr.cache, &entry) < 0) {
+			/*
+			 * Overflowed, but the global table cannot be expanded
+			 * because of deadlock.
+			 */
+			return -1;
+		}
+	}
+	return 0;
+}
+
+/**
+ * Look up address in the original global MR list.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param[out] entry
+ *   Pointer to returning MR cache entry. If no match, this will not be updated.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Found MR on match, NULL otherwise.
+ */
+static struct mlx4_mr *
+mr_lookup_dev_list(struct rte_eth_dev *dev, struct mlx4_mr_cache *entry,
+		   uintptr_t addr)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr *mr;
+
+	/* Iterate all the existing MRs. */
+	LIST_FOREACH(mr, &priv->mr.mr_list, mr) {
+		unsigned int n;
+
+		if (mr->ms_n == 0)
+			continue;
+		for (n = 0; n < mr->ms_bmp_n; ) {
+			struct mlx4_mr_cache ret = { 0, };
+
+			n = mr_find_next_chunk(mr, &ret, n);
+			if (addr >= ret.start && addr < ret.end) {
+				/* Found. */
+				*entry = ret;
+				return mr;
+			}
+		}
+	}
+	return NULL;
+}
+
+/**
+ * Look up address on device.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param[out] entry
+ *   Pointer to returning MR cache entry. If no match, this will not be updated.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on failure and rte_errno is set.
+ */
+static uint32_t
+mr_lookup_dev(struct rte_eth_dev *dev, struct mlx4_mr_cache *entry,
+	      uintptr_t addr)
+{
+	struct priv *priv = dev->data->dev_private;
+	uint16_t idx;
+	uint32_t lkey = UINT32_MAX;
+	struct mlx4_mr *mr;
+
+	/*
+	 * If the global cache has overflowed since it failed to expand the
+	 * B-tree table, it can't have all the exisitng MRs. Then, the address
+	 * has to be searched by traversing the original MR list instead, which
+	 * is very slow path. Otherwise, the global cache is all inclusive.
+	 */
+	if (!unlikely(priv->mr.cache.overflow)) {
+		lkey = mr_btree_lookup(&priv->mr.cache, &idx, addr);
+		if (lkey != UINT32_MAX)
+			*entry = (*priv->mr.cache.table)[idx];
+	} else {
+		/* Falling back to the slowest path. */
+		mr = mr_lookup_dev_list(dev, entry, addr);
+		if (mr != NULL)
+			lkey = entry->lkey;
+	}
+	assert(lkey == UINT32_MAX || (addr >= entry->start &&
+				      addr < entry->end));
+	return lkey;
+}
+
+/**
+ * Free MR resources. MR lock must not be held to avoid a deadlock. rte_free()
+ * can raise memory free event and the callback function will spin on the lock.
+ *
+ * @param mr
+ *   Pointer to MR to free.
+ */
+static void
+mr_free(struct mlx4_mr *mr)
+{
+	if (mr == NULL)
+		return;
+	DEBUG("freeing MR(%p):", (void *)mr);
+	if (mr->ibv_mr != NULL)
+		claim_zero(mlx4_glue->dereg_mr(mr->ibv_mr));
+	if (mr->ms_bmp != NULL)
+		rte_bitmap_free(mr->ms_bmp);
+	rte_free(mr);
+}
+
+/**
+ * Free Memory Region (MR).
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param mr
+ *   Pointer to MR to free.
+ */
+void
+mlx4_mr_free(struct rte_eth_dev *dev, struct mlx4_mr *mr)
+{
+	struct priv *priv = dev->data->dev_private;
+
+	/* Detach from the list and free resources later. */
+	rte_rwlock_write_lock(&priv->mr.rwlock);
+	LIST_REMOVE(mr, mr);
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+	/*
+	 * rte_free() inside can't be called with holding the lock. This could
+	 * cause deadlock when calling free callback.
+	 */
+	mr_free(mr);
+	DEBUG("port %u MR(%p) freed", dev->data->port_id, (void *)mr);
+}
+
+/**
+ * Releass resources of detached MR having no online entry.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ */
+static void
+mlx4_mr_garbage_collect(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr *mr_next;
+	struct mlx4_mr_list free_list = LIST_HEAD_INITIALIZER(free_list);
+
+	/*
+	 * MR can't be freed with holding the lock because rte_free() could call
+	 * memory free callback function. This will be a deadlock situation.
+	 */
+	rte_rwlock_write_lock(&priv->mr.rwlock);
+	/* Detach the whole free list and release it after unlocking. */
+	free_list = priv->mr.mr_free_list;
+	LIST_INIT(&priv->mr.mr_free_list);
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+	/* Release resources. */
+	mr_next = LIST_FIRST(&free_list);
+	while (mr_next != NULL) {
+		struct mlx4_mr *mr = mr_next;
+
+		mr_next = LIST_NEXT(mr, mr);
+		mr_free(mr);
+	}
+}
+
+/* Called during rte_memseg_contig_walk() by mlx4_mr_create(). */
+static int
+mr_find_contig_memsegs_cb(const struct rte_memseg_list *msl,
+			  const struct rte_memseg *ms, size_t len, void *arg)
+{
+	struct mr_find_contig_memsegs_data *data = arg;
+
+	if (data->addr < ms->addr_64 || data->addr >= ms->addr_64 + len)
+		return 0;
+	/* Found, save it and stop walking. */
+	data->start = ms->addr_64;
+	data->end = ms->addr_64 + len;
+	data->msl = msl;
+	return 1;
+}
+
+/**
+ * Create a new global Memroy Region (MR) for a missing virtual address.
+ * Register entire virtually contiguous memory chunk around the address.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param[out] entry
+ *   Pointer to returning MR cache entry, found in the global cache or newly
+ *   created. If failed to create one, this will not be updated.
+ * @param addr
+ *   Target virtual address to register.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on failure and rte_errno is set.
+ */
+static uint32_t
+mlx4_mr_create(struct rte_eth_dev *dev, struct mlx4_mr_cache *entry,
+	       uintptr_t addr)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct rte_mem_config *mcfg = rte_eal_get_configuration()->mem_config;
+	const struct rte_memseg_list *msl;
+	const struct rte_memseg *ms;
+	struct mlx4_mr *mr = NULL;
+	size_t len;
+	uint32_t ms_n;
+	uint32_t bmp_size;
+	void *bmp_mem;
+	int ms_idx_shift = -1;
+	unsigned int n;
+	struct mr_find_contig_memsegs_data data = {
+		.addr = addr,
+	};
+	struct mr_find_contig_memsegs_data data_re;
+
+	DEBUG("port %u creating a MR using address (%p)",
+	      dev->data->port_id, (void *)addr);
+	/*
+	 * Release detached MRs if any. This can't be called with holding either
+	 * memory_hotplug_lock or priv->mr.rwlock. MRs on the free list have
+	 * been detached by the memory free event but it couldn't be released
+	 * inside the callback due to deadlock. As a result, releasing resources
+	 * is quite opportunistic.
+	 */
+	mlx4_mr_garbage_collect(dev);
+	/*
+	 * Find out a contiguous virtual address chunk in use, to which the
+	 * given address belongs, in order to register maximum range. In the
+	 * best case where mempools are not dynamically recreated and
+	 * '--socket-mem' is speicified as an EAL option, it is very likely to
+	 * have only one MR(LKey) per a socket and per a hugepage-size even
+	 * though the system memory is highly fragmented.
+	 */
+	if (!rte_memseg_contig_walk(mr_find_contig_memsegs_cb, &data)) {
+		WARN("port %u unable to find virtually contigous"
+		     " chunk for address (%p)."
+		     " rte_memseg_contig_walk() failed.",
+		     dev->data->port_id, (void *)addr);
+		rte_errno = ENXIO;
+		goto err_nolock;
+	}
+alloc_resources:
+	/* Addresses must be page-aligned. */
+	assert(rte_is_aligned((void *)data.start, data.msl->page_sz));
+	assert(rte_is_aligned((void *)data.end, data.msl->page_sz));
+	msl = data.msl;
+	ms = rte_mem_virt2memseg((void *)data.start, msl);
+	len = data.end - data.start;
+	assert(msl->page_sz == ms->hugepage_sz);
+	/* Number of memsegs in the range. */
+	ms_n = len / msl->page_sz;
+	DEBUG("port %u extending %p to [0x%lx, 0x%lx), page_sz=0x%lx, ms_n=%u",
+	      dev->data->port_id, (void *)addr,
+	      data.start, data.end, msl->page_sz, ms_n);
+	/* Size of memory for bitmap. */
+	bmp_size = rte_bitmap_get_memory_footprint(ms_n);
+	mr = rte_zmalloc_socket(NULL,
+				RTE_ALIGN_CEIL(sizeof(*mr),
+					       RTE_CACHE_LINE_SIZE) +
+				bmp_size,
+				RTE_CACHE_LINE_SIZE, msl->socket_id);
+	if (mr == NULL) {
+		WARN("port %u unable to allocate memory for a new MR of"
+		     " address (%p).",
+		     dev->data->port_id, (void *)addr);
+		rte_errno = ENOMEM;
+		goto err_nolock;
+	}
+	mr->msl = msl;
+	/*
+	 * Save the index of the first memseg and initialize memseg bitmap. To
+	 * see if a memseg of ms_idx in the memseg-list is still valid, check:
+	 *	rte_bitmap_get(mr->bmp, ms_idx - mr->ms_base_idx)
+	 */
+	mr->ms_base_idx = rte_fbarray_find_idx(&msl->memseg_arr, ms);
+	bmp_mem = RTE_PTR_ALIGN_CEIL(mr + 1, RTE_CACHE_LINE_SIZE);
+	mr->ms_bmp = rte_bitmap_init(ms_n, bmp_mem, bmp_size);
+	if (mr->ms_bmp == NULL) {
+		WARN("port %u unable to initialize bitamp for a new MR of"
+		     " address (%p).",
+		     dev->data->port_id, (void *)addr);
+		rte_errno = EINVAL;
+		goto err_nolock;
+	}
+	/*
+	 * Should recheck whether the extended contiguous chunk is still valid.
+	 * Because memory_hotplug_lock can't be held if there's any memory
+	 * related calls in a critical path, resource allocation above can't be
+	 * locked. If the memory has been changed at this point, try again with
+	 * just single page. If not, go on with the big chunk atomically from
+	 * here.
+	 */
+	rte_rwlock_read_lock(&mcfg->memory_hotplug_lock);
+	data_re = data;
+	if (len > msl->page_sz &&
+	    !rte_memseg_contig_walk(mr_find_contig_memsegs_cb, &data_re)) {
+		WARN("port %u unable to find virtually contigous"
+		     " chunk for address (%p)."
+		     " rte_memseg_contig_walk() failed.",
+		     dev->data->port_id, (void *)addr);
+		rte_errno = ENXIO;
+		goto err_memlock;
+	}
+	if (data.start != data_re.start || data.end != data_re.end) {
+		/*
+		 * The extended contiguous chunk has been changed. Try again
+		 * with single memseg instead.
+		 */
+		data.start = RTE_ALIGN_FLOOR(addr, msl->page_sz);
+		data.end = data.start + msl->page_sz;
+		rte_rwlock_read_unlock(&mcfg->memory_hotplug_lock);
+		mr_free(mr);
+		goto alloc_resources;
+	}
+	assert(data.msl == data_re.msl);
+	rte_rwlock_write_lock(&priv->mr.rwlock);
+	/*
+	 * Check the address is really missing. If other thread already created
+	 * one or it is not found due to overflow, abort and return.
+	 */
+	if (mr_lookup_dev(dev, entry, addr) != UINT32_MAX) {
+		/*
+		 * Insert to the global cache table. It may fail due to
+		 * low-on-memory. Then, this entry will have to be searched
+		 * here again.
+		 */
+		mr_btree_insert(&priv->mr.cache, entry);
+		DEBUG("port %u found MR for %p on final lookup, abort",
+		      dev->data->port_id, (void *)addr);
+		rte_rwlock_write_unlock(&priv->mr.rwlock);
+		rte_rwlock_read_unlock(&mcfg->memory_hotplug_lock);
+		/*
+		 * Must be unlocked before calling rte_free() because
+		 * mlx4_mr_mem_event_free_cb() can be called inside.
+		 */
+		mr_free(mr);
+		return entry->lkey;
+	}
+	/*
+	 * Trim start and end addresses for verbs MR. Set bits for registering
+	 * memsegs but exclude already registered ones. Bitmap can be
+	 * fragmented.
+	 */
+	for (n = 0; n < ms_n; ++n) {
+		uintptr_t start;
+		struct mlx4_mr_cache ret = { 0, };
+
+		start = data_re.start + n * msl->page_sz;
+		/* Exclude memsegs already registered by other MRs. */
+		if (mr_lookup_dev(dev, &ret, start) == UINT32_MAX) {
+			/*
+			 * Start from the first unregistered memseg in the
+			 * extended range.
+			 */
+			if (ms_idx_shift == -1) {
+				mr->ms_base_idx += n;
+				data.start = start;
+				ms_idx_shift = n;
+			}
+			data.end = start + msl->page_sz;
+			rte_bitmap_set(mr->ms_bmp, n - ms_idx_shift);
+			++mr->ms_n;
+		}
+	}
+	len = data.end - data.start;
+	mr->ms_bmp_n = len / msl->page_sz;
+	assert(ms_idx_shift + mr->ms_bmp_n <= ms_n);
+	/*
+	 * Finally create a verbs MR for the memory chunk. ibv_reg_mr() can be
+	 * called with holding the memory lock because it doesn't use
+	 * mlx4_alloc_buf_extern() which eventually calls rte_malloc_socket()
+	 * through mlx4_alloc_verbs_buf().
+	 */
+	mr->ibv_mr = mlx4_glue->reg_mr(priv->pd, (void *)data.start, len,
+				       IBV_ACCESS_LOCAL_WRITE);
+	if (mr->ibv_mr == NULL) {
+		WARN("port %u fail to create a verbs MR for address (%p)",
+		     dev->data->port_id, (void *)addr);
+		rte_errno = EINVAL;
+		goto err_mrlock;
+	}
+	assert((uintptr_t)mr->ibv_mr->addr == data.start);
+	assert(mr->ibv_mr->length == len);
+	LIST_INSERT_HEAD(&priv->mr.mr_list, mr, mr);
+	DEBUG("port %u MR CREATED (%p) for %p:\n"
+	      "  [0x%lx, 0x%lx), lkey=0x%x base_idx=%u ms_n=%u, ms_bmp_n=%u",
+	      dev->data->port_id, (void *)mr, (void *)addr,
+	      data.start, data.end, rte_cpu_to_be_32(mr->ibv_mr->lkey),
+	      mr->ms_base_idx, mr->ms_n, mr->ms_bmp_n);
+	/* Insert to the global cache table. */
+	mr_insert_dev_cache(dev, mr);
+	/* Fill in output data. */
+	mr_lookup_dev(dev, entry, addr);
+	/* Lookup can't fail. */
+	assert(entry->lkey != UINT32_MAX);
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+	rte_rwlock_read_unlock(&mcfg->memory_hotplug_lock);
+	return entry->lkey;
+err_mrlock:
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+err_memlock:
+	rte_rwlock_read_unlock(&mcfg->memory_hotplug_lock);
+err_nolock:
+	/*
+	 * In case of error, as this can be called in a datapath, a warning
+	 * message per an error is preferable instead. Must be unlocked before
+	 * calling rte_free() because mlx4_mr_mem_event_free_cb() can be called
+	 * inside.
+	 */
+	mr_free(mr);
+	return UINT32_MAX;
+}
+
+/**
+ * Rebuild the global B-tree cache of device from the original MR list.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ */
+static void
+mr_rebuild_dev_cache(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr *mr;
+
+	DEBUG("port %u rebuild dev cache[]", dev->data->port_id);
+	/* Flush cache to rebuild. */
+	priv->mr.cache.len = 1;
+	priv->mr.cache.overflow = 0;
+	/* Iterate all the existing MRs. */
+	LIST_FOREACH(mr, &priv->mr.mr_list, mr)
+		if (mr_insert_dev_cache(dev, mr) < 0)
+			return;
+}
+
+/**
+ * Callback for memory free event. Iterate freed memsegs and check whether it
+ * belongs to an existing MR. If found, clear the bit from bitmap of MR. As a
+ * result, the MR would be fragmented. If it becomes empty, the MR will be freed
+ * later by mlx4_mr_garbage_collect().
+ *
+ * The global cache must be rebuilt if there's any change and this event has to
+ * be propagated to dataplane threads to flush the local caches.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param addr
+ *   Address of freed memory.
+ * @param len
+ *   Size of freed memory.
+ */
+static void
+mlx4_mr_mem_event_free_cb(struct rte_eth_dev *dev, const void *addr, size_t len)
+{
+	struct priv *priv = dev->data->dev_private;
+	const struct rte_memseg_list *msl;
+	struct mlx4_mr *mr;
+	int ms_n;
+	int i;
+	int rebuild = 0;
+
+	DEBUG("port %u free callback: addr=%p, len=%lu",
+	      dev->data->port_id, addr, len);
+	msl = rte_mem_virt2memseg_list(addr);
+	/* addr and len must be page-aligned. */
+	assert((uintptr_t)addr == RTE_ALIGN((uintptr_t)addr, msl->page_sz));
+	assert(len == RTE_ALIGN(len, msl->page_sz));
+	ms_n = len / msl->page_sz;
+	rte_rwlock_write_lock(&priv->mr.rwlock);
+	/* Clear bits of freed memsegs from MR. */
+	for (i = 0; i < ms_n; ++i) {
+		const struct rte_memseg *ms;
+		struct mlx4_mr_cache entry;
+		uintptr_t start;
+		int ms_idx;
+		uint32_t pos;
+
+		/* Find MR having this memseg. */
+		start = (uintptr_t)addr + i * msl->page_sz;
+		mr = mr_lookup_dev_list(dev, &entry, start);
+		if (mr == NULL)
+			continue;
+		ms = rte_mem_virt2memseg((void *)start, msl);
+		assert(ms != NULL);
+		assert(msl->page_sz == ms->hugepage_sz);
+		ms_idx = rte_fbarray_find_idx(&msl->memseg_arr, ms);
+		pos = ms_idx - mr->ms_base_idx;
+		assert(rte_bitmap_get(mr->ms_bmp, pos));
+		assert(pos < mr->ms_bmp_n);
+		DEBUG("port %u MR(%p): clear bitmap[%u] for addr %p",
+		      dev->data->port_id, (void *)mr, pos, (void *)start);
+		rte_bitmap_clear(mr->ms_bmp, pos);
+		if (--mr->ms_n == 0) {
+			LIST_REMOVE(mr, mr);
+			LIST_INSERT_HEAD(&priv->mr.mr_free_list, mr, mr);
+			DEBUG("port %u remove MR(%p) from list",
+			      dev->data->port_id, (void *)mr);
+		}
+		/*
+		 * MR is fragmented or will be freed. the global cache must be
+		 * rebuilt.
+		 */
+		rebuild = 1;
+	}
+	if (rebuild) {
+		mr_rebuild_dev_cache(dev);
+		/*
+		 * Flush local caches by propagating invalidation across cores.
+		 * rte_smp_wmb() is enough to synchronize this event. If one of
+		 * freed memsegs is seen by other core, that means the memseg
+		 * has been allocated by allocator, which will come after this
+		 * free call. Therefore, this store instruction (incrementing
+		 * generation below) will be guaranteed to be seen by other core
+		 * before the core sees the newly allocated memory.
+		 */
+		++priv->mr.dev_gen;
+		DEBUG("broadcasting local cache flush, gen=%d",
+		      priv->mr.dev_gen);
+		rte_smp_wmb();
+	}
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+#ifndef NDEBUG
+	if (rebuild)
+		mlx4_mr_dump_dev(dev);
+#endif
+}
+
+/**
+ * Callback for memory event.
+ *
+ * @param event_type
+ *   Memory event type.
+ * @param addr
+ *   Address of memory.
+ * @param len
+ *   Size of memory.
+ */
+void
+mlx4_mr_mem_event_cb(enum rte_mem_event event_type, const void *addr,
+		     size_t len)
+{
+	struct priv *priv;
+
+	switch (event_type) {
+	case RTE_MEM_EVENT_FREE:
+		rte_rwlock_read_lock(&mlx4_mem_event_rwlock);
+		/* Iterate all the existing mlx4 devices. */
+		LIST_FOREACH(priv, &mlx4_mem_event_cb_list, mem_event_cb)
+			mlx4_mr_mem_event_free_cb(priv->dev, addr, len);
+		rte_rwlock_read_unlock(&mlx4_mem_event_rwlock);
+		break;
+	case RTE_MEM_EVENT_ALLOC:
+	default:
+		break;
+	}
+}
+
+/**
+ * Look up address in the global MR cache table. If not found, create a new MR.
+ * Insert the found/created entry to local bottom-half cache table.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param mr_ctrl
+ *   Pointer to per-queue MR control structure.
+ * @param[out] entry
+ *   Pointer to returning MR cache entry, found in the global cache or newly
+ *   created. If failed to create one, this is not written.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static uint32_t
+mlx4_mr_lookup_dev(struct rte_eth_dev *dev, struct mlx4_mr_ctrl *mr_ctrl,
+		   struct mlx4_mr_cache *entry, uintptr_t addr)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr_btree *bt = &mr_ctrl->cache_bh;
+	uint16_t idx;
+	uint32_t lkey;
+
+	/* If local cache table is full, try to double it. */
+	if (unlikely(bt->len == bt->size))
+		mr_btree_expand(bt, bt->size << 1);
+	/* Look up in the global cache. */
+	rte_rwlock_read_lock(&priv->mr.rwlock);
+	lkey = mr_btree_lookup(&priv->mr.cache, &idx, addr);
+	if (lkey != UINT32_MAX) {
+		/* Found. */
+		*entry = (*priv->mr.cache.table)[idx];
+		rte_rwlock_read_unlock(&priv->mr.rwlock);
+		/*
+		 * Update local cache. Even if it fails, return the found entry
+		 * to update top-half cache. Next time, this entry will be found
+		 * in the global cache.
+		 */
+		mr_btree_insert(bt, entry);
+		return lkey;
+	}
+	rte_rwlock_read_unlock(&priv->mr.rwlock);
+	/* First time to see the address? Create a new MR. */
+	lkey = mlx4_mr_create(dev, entry, addr);
+	/*
+	 * Update the local cache if successfully created a new global MR. Even
+	 * if failed to create one, there's no action to take in this datapath
+	 * code. As returning LKey is invalid, this will eventually make HW
+	 * fail.
+	 */
+	if (lkey != UINT32_MAX)
+		mr_btree_insert(bt, entry);
+	return lkey;
+}
+
+/**
+ * Bottom-half of LKey search on datapath. Firstly search in cache_bh[] and if
+ * misses, search in the global MR cache table and update the new entry to
+ * per-queue local caches.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param mr_ctrl
+ *   Pointer to per-queue MR control structure.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static uint32_t
+mlx4_mr_addr2mr_bh(struct rte_eth_dev *dev, struct mlx4_mr_ctrl *mr_ctrl,
+		   uintptr_t addr)
+{
+	uint32_t lkey;
+	uint16_t bh_idx = 0;
+	/* Victim in top-half cache to replace with new entry. */
+	struct mlx4_mr_cache *repl = &mr_ctrl->cache[mr_ctrl->head];
+
+	/* Binary-search MR translation table. */
+	lkey = mr_btree_lookup(&mr_ctrl->cache_bh, &bh_idx, addr);
+	/* Update top-half cache. */
+	if (likely(lkey != UINT32_MAX)) {
+		*repl = (*mr_ctrl->cache_bh.table)[bh_idx];
+	} else {
+		/*
+		 * If missed in local lookup table, search in the global cache
+		 * and local cache_bh[] will be updated inside if possible.
+		 * Top-half cache entry will also be updated.
+		 */
+		lkey = mlx4_mr_lookup_dev(dev, mr_ctrl, repl, addr);
+		if (unlikely(lkey == UINT32_MAX))
+			return UINT32_MAX;
+	}
+	/* Update the most recently used entry. */
+	mr_ctrl->mru = mr_ctrl->head;
+	/* Point to the next victim, the oldest. */
+	mr_ctrl->head = (mr_ctrl->head + 1) % MLX4_MR_CACHE_N;
+	return lkey;
+}
+
+/**
+ * Bottom-half of LKey search on Rx.
+ *
+ * @param rxq
+ *   Pointer to Rx queue structure.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+uint32_t
+mlx4_rx_addr2mr_bh(struct rxq *rxq, uintptr_t addr)
+{
+	struct mlx4_mr_ctrl *mr_ctrl = &rxq->mr_ctrl;
+	struct priv *priv = rxq->priv;
+
+	DEBUG("Rx queue %u: miss on top-half, mru=%u, head=%u, addr=%p",
+	      rxq->stats.idx, mr_ctrl->mru, mr_ctrl->head, (void *)addr);
+	return mlx4_mr_addr2mr_bh(priv->dev, mr_ctrl, addr);
+}
+
+/**
+ * Bottom-half of LKey search on Tx.
+ *
+ * @param txq
+ *   Pointer to Tx queue structure.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+uint32_t
+mlx4_tx_addr2mr_bh(struct txq *txq, uintptr_t addr)
+{
+	struct mlx4_mr_ctrl *mr_ctrl = &txq->mr_ctrl;
+	struct priv *priv = txq->priv;
+
+	DEBUG("Tx queue %u: miss on top-half, mru=%u, head=%u, addr=%p",
+	      txq->stats.idx, mr_ctrl->mru, mr_ctrl->head, (void *)addr);
+	return mlx4_mr_addr2mr_bh(priv->dev, mr_ctrl, addr);
+}
+
+/**
+ * Flush all of the local cache entries.
+ *
+ * @param mr_ctrl
+ *   Pointer to per-queue MR control structure.
+ */
+void
+mlx4_mr_flush_local_cache(struct mlx4_mr_ctrl *mr_ctrl)
+{
+	/* Reset the most-recently-used index. */
+	mr_ctrl->mru = 0;
+	/* Reset the linear search array. */
+	mr_ctrl->head = 0;
+	memset(mr_ctrl->cache, 0, sizeof(mr_ctrl->cache));
+	/* Reset the B-tree table. */
+	mr_ctrl->cache_bh.len = 1;
+	mr_ctrl->cache_bh.overflow = 0;
+	/* Update the generation number. */
+	mr_ctrl->cur_gen = *mr_ctrl->dev_gen_ptr;
+	DEBUG("mr_ctrl(%p): flushed, cur_gen=%d",
+	      (void *)mr_ctrl, mr_ctrl->cur_gen);
+}
+
+/* Called during rte_mempool_mem_iter() by mlx4_mr_update_mp(). */
+static void
+mlx4_mr_update_mp_cb(struct rte_mempool *mp __rte_unused, void *opaque,
+		     struct rte_mempool_memhdr *memhdr,
+		     unsigned mem_idx __rte_unused)
+{
+	struct mr_update_mp_data *data = opaque;
+	uint32_t lkey;
+
+	/* Stop iteration if failed in the previous walk. */
+	if (data->ret < 0)
+		return;
+	/* Register address of the chunk and update local caches. */
+	lkey = mlx4_mr_addr2mr_bh(data->dev, data->mr_ctrl,
+				  (uintptr_t)memhdr->addr);
+	if (lkey == UINT32_MAX)
+		data->ret = -1;
+}
+
+/**
+ * Register entire memory chunks in a Mempool.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ * @param mr_ctrl
+ *   Pointer to per-queue MR control structure.
+ * @param mp
+ *   Pointer to registering Mempool.
+ *
+ * @return
+ *   0 on success, -1 on failure.
+ */
+int
+mlx4_mr_update_mp(struct rte_eth_dev *dev, struct mlx4_mr_ctrl *mr_ctrl,
+		  struct rte_mempool *mp)
+{
+	struct mr_update_mp_data data = {
+		.dev = dev,
+		.mr_ctrl = mr_ctrl,
+		.ret = 0,
+	};
+
+	rte_mempool_mem_iter(mp, mlx4_mr_update_mp_cb, &data);
+	return data.ret;
+}
+
+#ifndef NDEBUG
+/**
+ * Dump all the created MRs and the global cache entries.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ */
+void
+mlx4_mr_dump_dev(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr *mr;
+	int mr_n = 0;
+	int chunk_n = 0;
+
+	rte_rwlock_read_lock(&priv->mr.rwlock);
+	/* Iterate all the existing MRs. */
+	LIST_FOREACH(mr, &priv->mr.mr_list, mr) {
+		unsigned int n;
+
+		DEBUG("port %u MR[%u], LKey = 0x%x, ms_n = %u, ms_bmp_n = %u",
+		      dev->data->port_id, mr_n++,
+		      rte_cpu_to_be_32(mr->ibv_mr->lkey),
+		      mr->ms_n, mr->ms_bmp_n);
+		if (mr->ms_n == 0)
+			continue;
+		for (n = 0; n < mr->ms_bmp_n; ) {
+			struct mlx4_mr_cache ret = { 0, };
+
+			n = mr_find_next_chunk(mr, &ret, n);
+			if (!ret.end)
+				break;
+			DEBUG("  chunk[%u], [0x%lx, 0x%lx)",
+			      chunk_n++, ret.start, ret.end);
+		}
+	}
+	DEBUG("port %u dumping global cache", dev->data->port_id);
+	mlx4_mr_btree_dump(&priv->mr.cache);
+	rte_rwlock_read_unlock(&priv->mr.rwlock);
+}
+#endif
+
+/**
+ * Release all the created MRs and resources. Remove device from memory callback
+ * list.
+ *
+ * @param dev
+ *   Pointer to Ethernet device.
+ */
+void
+mlx4_mr_release(struct rte_eth_dev *dev)
+{
+	struct priv *priv = dev->data->dev_private;
+	struct mlx4_mr *mr_next = LIST_FIRST(&priv->mr.mr_list);
+
+	/* Remove from memory callback device list. */
+	rte_rwlock_write_lock(&mlx4_mem_event_rwlock);
+	LIST_REMOVE(priv, mem_event_cb);
+	rte_rwlock_write_unlock(&mlx4_mem_event_rwlock);
+#ifndef NDEBUG
+	mlx4_mr_dump_dev(dev);
+#endif
+	rte_rwlock_write_lock(&priv->mr.rwlock);
+	/* Detach from MR list and move to free list. */
+	while (mr_next != NULL) {
+		struct mlx4_mr *mr = mr_next;
+
+		mr_next = LIST_NEXT(mr, mr);
+		LIST_REMOVE(mr, mr);
+		LIST_INSERT_HEAD(&priv->mr.mr_free_list, mr, mr);
+	}
+	LIST_INIT(&priv->mr.mr_list);
+	/* Free global cache. */
+	mlx4_mr_btree_free(&priv->mr.cache);
+	rte_rwlock_write_unlock(&priv->mr.rwlock);
+	/* Free all remaining MRs. */
+	mlx4_mr_garbage_collect(dev);
+}
diff --git a/drivers/net/mlx4/mlx4_mr.h b/drivers/net/mlx4/mlx4_mr.h
new file mode 100644
index 000000000..33305c0f8
--- /dev/null
+++ b/drivers/net/mlx4/mlx4_mr.h
@@ -0,0 +1,126 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright 2018 6WIND S.A.
+ * Copyright 2018 Mellanox Technologies, Ltd
+ */
+
+#ifndef RTE_PMD_MLX4_MR_H_
+#define RTE_PMD_MLX4_MR_H_
+
+#include <stddef.h>
+#include <stdint.h>
+#include <sys/queue.h>
+
+/* Verbs headers do not support -pedantic. */
+#ifdef PEDANTIC
+#pragma GCC diagnostic ignored "-Wpedantic"
+#endif
+#include <infiniband/verbs.h>
+#ifdef PEDANTIC
+#pragma GCC diagnostic error "-Wpedantic"
+#endif
+
+#include <rte_eal_memconfig.h>
+#include <rte_ethdev.h>
+#include <rte_rwlock.h>
+#include <rte_bitmap.h>
+
+/* Size of per-queue MR cache array for linear search. */
+#define MLX4_MR_CACHE_N 8
+
+/* Size of MR cache table for binary search. */
+#define MLX4_MR_BTREE_CACHE_N 256
+
+/* Memory Region object. */
+struct mlx4_mr {
+	LIST_ENTRY(mlx4_mr) mr; /**< Pointer to the prev/next entry. */
+	struct ibv_mr *ibv_mr; /* Verbs Memory Region. */
+	const struct rte_memseg_list *msl;
+	int ms_base_idx; /* Start index of msl->memseg_arr[]. */
+	int ms_n; /* Number of memsegs in use. */
+	uint32_t ms_bmp_n; /* Number of bits in memsegs bit-mask. */
+	struct rte_bitmap *ms_bmp; /* Bit-mask of memsegs belonged to MR. */
+};
+
+/* Cache entry for Memory Region. */
+struct mlx4_mr_cache {
+	uintptr_t start; /* Start address of MR. */
+	uintptr_t end; /* End address of MR. */
+	uint32_t lkey; /* rte_cpu_to_be_32(ibv_mr->lkey). */
+} __rte_packed;
+
+/* MR Cache table for Binary search. */
+struct mlx4_mr_btree {
+	uint16_t len; /* Number of entries. */
+	uint16_t size; /* Total number of entries. */
+	int overflow; /* Mark failure of table expansion. */
+	struct mlx4_mr_cache (*table)[];
+} __rte_packed;
+
+/* Per-queue MR control descriptor. */
+struct mlx4_mr_ctrl {
+	uint32_t *dev_gen_ptr; /* Generation number of device to poll. */
+	uint32_t cur_gen; /* Generation number saved to flush caches. */
+	uint16_t mru; /* Index of last hit entry in top-half cache. */
+	uint16_t head; /* Index of the oldest entry in top-half cache. */
+	struct mlx4_mr_cache cache[MLX4_MR_CACHE_N]; /* Cache for top-half. */
+	struct mlx4_mr_btree cache_bh; /* Cache for bottom-half. */
+} __rte_packed;
+
+/* First entry must be NULL for comparison. */
+#define MR_N(n) ((n) - 1)
+
+/* Whether there's only one entry in MR lookup table. */
+#define IS_SINGLE_MR(n) (MR_N(n) == 1)
+
+extern struct mlx4_dev_list  mlx4_mem_event_cb_list;
+extern rte_rwlock_t mlx4_mem_event_rwlock;
+
+void mlx4_mr_free(struct rte_eth_dev *dev, struct mlx4_mr *mr);
+int mlx4_mr_btree_init(struct mlx4_mr_btree *bt, int n, int socket);
+void mlx4_mr_btree_free(struct mlx4_mr_btree *bt);
+void mlx4_mr_btree_dump(struct mlx4_mr_btree *bt);
+void mlx4_mr_mem_event_cb(enum rte_mem_event event_type, const void *addr,
+			  size_t len);
+int mlx4_mr_update_mp(struct rte_eth_dev *dev, struct mlx4_mr_ctrl *mr_ctrl,
+		      struct rte_mempool *mp);
+void mlx4_mr_dump_dev(struct rte_eth_dev *dev);
+void mlx4_mr_release(struct rte_eth_dev *dev);
+
+/**
+ * Look up LKey from given lookup table by linear search. Firstly look up the
+ * last-hit entry. If miss, the entire array is searched. If found, update the
+ * last-hit index and return LKey.
+ *
+ * @param lkp_tbl
+ *   Pointer to lookup table.
+ * @param[in,out] cached_idx
+ *   Pointer to last-hit index.
+ * @param n
+ *   Size of lookup table.
+ * @param addr
+ *   Search key.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static __rte_always_inline uint32_t
+mlx4_mr_lookup_cache(struct mlx4_mr_cache *lkp_tbl, uint16_t *cached_idx,
+		     uint16_t n, uintptr_t addr)
+{
+	uint16_t idx;
+
+	if (likely(addr >= lkp_tbl[*cached_idx].start &&
+		   addr < lkp_tbl[*cached_idx].end))
+		return lkp_tbl[*cached_idx].lkey;
+	for (idx = 0; idx < n && lkp_tbl[idx].start != 0; ++idx) {
+		if (addr >= lkp_tbl[idx].start &&
+		    addr < lkp_tbl[idx].end) {
+			/* Found. */
+			*cached_idx = idx;
+			return lkp_tbl[idx].lkey;
+		}
+	}
+	return UINT32_MAX;
+}
+
+#endif /* RTE_PMD_MLX4_MR_H_ */
diff --git a/drivers/net/mlx4/mlx4_rxq.c b/drivers/net/mlx4/mlx4_rxq.c
index 5621d5bd4..ad706be82 100644
--- a/drivers/net/mlx4/mlx4_rxq.c
+++ b/drivers/net/mlx4/mlx4_rxq.c
@@ -488,6 +488,7 @@ mlx4_rxq_attach(struct rxq *rxq)
 	}
 
 	struct priv *priv = rxq->priv;
+	struct rte_eth_dev *dev = priv->dev;
 	const uint32_t elts_n = 1 << rxq->elts_n;
 	const uint32_t sges_n = 1 << rxq->sges_n;
 	struct rte_mbuf *(*elts)[elts_n] = rxq->elts;
@@ -552,6 +553,11 @@ mlx4_rxq_attach(struct rxq *rxq)
 		msg = "failed to obtain device information from WQ/CQ objects";
 		goto error;
 	}
+	/* Pre-register Rx mempool. */
+	DEBUG("port %u Rx queue %u registering mp %s having %u chunks",
+	      priv->dev->data->port_id, rxq->stats.idx,
+	      rxq->mp->name, rxq->mp->nb_mem_chunks);
+	mlx4_mr_update_mp(dev, &rxq->mr_ctrl, rxq->mp);
 	wqes = (volatile struct mlx4_wqe_data_seg (*)[])
 		((uintptr_t)dv_rwq.buf.buf + dv_rwq.rq.offset);
 	for (i = 0; i != RTE_DIM(*elts); ++i) {
@@ -583,7 +589,7 @@ mlx4_rxq_attach(struct rxq *rxq)
 			.addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(buf,
 								  uintptr_t)),
 			.byte_count = rte_cpu_to_be_32(buf->data_len),
-			.lkey = UINT32_MAX,
+			.lkey = mlx4_rx_mb2mr(rxq, buf),
 		};
 		(*elts)[i] = buf;
 	}
@@ -883,6 +889,11 @@ mlx4_rx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
 		      1 << rxq->sges_n);
 		goto error;
 	}
+	if (mlx4_mr_btree_init(&rxq->mr_ctrl.cache_bh,
+			       MLX4_MR_BTREE_CACHE_N, socket)) {
+		/* rte_errno is already set. */
+		goto error;
+	}
 	if (dev->data->dev_conf.intr_conf.rxq) {
 		rxq->channel = mlx4_glue->create_comp_channel(priv->ctx);
 		if (rxq->channel == NULL) {
@@ -940,5 +951,6 @@ mlx4_rx_queue_release(void *dpdk_rxq)
 	assert(!rxq->rq_db);
 	if (rxq->channel)
 		claim_zero(mlx4_glue->destroy_comp_channel(rxq->channel));
+	mlx4_mr_btree_free(&rxq->mr_ctrl.cache_bh);
 	rte_free(rxq);
 }
diff --git a/drivers/net/mlx4/mlx4_rxtx.c b/drivers/net/mlx4/mlx4_rxtx.c
index 21ffd435a..f3899c79a 100644
--- a/drivers/net/mlx4/mlx4_rxtx.c
+++ b/drivers/net/mlx4/mlx4_rxtx.c
@@ -344,24 +344,6 @@ mlx4_txq_complete(struct txq *txq, const unsigned int elts_m,
 }
 
 /**
- * Get memory pool (MP) from mbuf. If mbuf is indirect, the pool from which
- * the cloned mbuf is allocated is returned instead.
- *
- * @param buf
- *   Pointer to mbuf.
- *
- * @return
- *   Memory pool where data is located for given mbuf.
- */
-static struct rte_mempool *
-mlx4_txq_mb2mp(struct rte_mbuf *buf)
-{
-	if (unlikely(RTE_MBUF_INDIRECT(buf)))
-		return rte_mbuf_from_indirect(buf)->pool;
-	return buf->pool;
-}
-
-/**
  * Write Tx data segment to the SQ.
  *
  * @param dseg
@@ -378,7 +360,7 @@ mlx4_fill_tx_data_seg(volatile struct mlx4_wqe_data_seg *dseg,
 		       uint32_t lkey, uintptr_t addr, rte_be32_t  byte_count)
 {
 	dseg->addr = rte_cpu_to_be_64(addr);
-	dseg->lkey = rte_cpu_to_be_32(lkey);
+	dseg->lkey = lkey;
 #if RTE_CACHE_LINE_SIZE < 64
 	/*
 	 * Need a barrier here before writing the byte_count
@@ -437,7 +419,7 @@ mlx4_tx_burst_segs(struct rte_mbuf *buf, struct txq *txq,
 	goto txbb_tail_segs;
 txbb_head_seg:
 	/* Memory region key (big endian) for this memory pool. */
-	lkey = mlx4_txq_mp2mr(txq, mlx4_txq_mb2mp(sbuf));
+	lkey = mlx4_tx_mb2mr(txq, sbuf);
 	if (unlikely(lkey == (uint32_t)-1)) {
 		DEBUG("%p: unable to get MP <-> MR association",
 		      (void *)txq);
@@ -449,7 +431,7 @@ mlx4_tx_burst_segs(struct rte_mbuf *buf, struct txq *txq,
 		dseg = (volatile struct mlx4_wqe_data_seg *)
 			sq->buf;
 	dseg->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(sbuf, uintptr_t));
-	dseg->lkey = rte_cpu_to_be_32(lkey);
+	dseg->lkey = lkey;
 	/*
 	 * This data segment starts at the beginning of a new
 	 * TXBB, so we need to postpone its byte_count writing
@@ -469,7 +451,7 @@ mlx4_tx_burst_segs(struct rte_mbuf *buf, struct txq *txq,
 	/* Jump to default if there are more than two segments remaining. */
 	switch (nb_segs) {
 	default:
-		lkey = mlx4_txq_mp2mr(txq, mlx4_txq_mb2mp(sbuf));
+		lkey = mlx4_tx_mb2mr(txq, sbuf);
 		if (unlikely(lkey == (uint32_t)-1)) {
 			DEBUG("%p: unable to get MP <-> MR association",
 			      (void *)txq);
@@ -485,7 +467,7 @@ mlx4_tx_burst_segs(struct rte_mbuf *buf, struct txq *txq,
 		nb_segs--;
 		/* fallthrough */
 	case 2:
-		lkey = mlx4_txq_mp2mr(txq, mlx4_txq_mb2mp(sbuf));
+		lkey = mlx4_tx_mb2mr(txq, sbuf);
 		if (unlikely(lkey == (uint32_t)-1)) {
 			DEBUG("%p: unable to get MP <-> MR association",
 			      (void *)txq);
@@ -501,7 +483,7 @@ mlx4_tx_burst_segs(struct rte_mbuf *buf, struct txq *txq,
 		nb_segs--;
 		/* fallthrough */
 	case 1:
-		lkey = mlx4_txq_mp2mr(txq, mlx4_txq_mb2mp(sbuf));
+		lkey = mlx4_tx_mb2mr(txq, sbuf);
 		if (unlikely(lkey == (uint32_t)-1)) {
 			DEBUG("%p: unable to get MP <-> MR association",
 			      (void *)txq);
@@ -611,7 +593,7 @@ mlx4_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
 				elt->buf = NULL;
 				break;
 			}
-			lkey = mlx4_txq_mp2mr(txq, mlx4_txq_mb2mp(buf));
+			lkey = mlx4_tx_mb2mr(txq, buf);
 			if (unlikely(lkey == (uint32_t)-1)) {
 				/* MR does not exist. */
 				DEBUG("%p: unable to get MP <-> MR association",
@@ -966,6 +948,9 @@ mlx4_rx_burst(void *dpdk_rxq, struct rte_mbuf **pkts, uint16_t pkts_n)
 		 * changes.
 		 */
 		scat->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t));
+		/* If there's only one MR, no need to replace LKey in WQE. */
+		if (unlikely(!IS_SINGLE_MR(rxq->mr_ctrl.cache_bh.len)))
+			scat->lkey = mlx4_rx_mb2mr(rxq, rep);
 		if (len > seg->data_len) {
 			len -= seg->data_len;
 			++pkt->nb_segs;
diff --git a/drivers/net/mlx4/mlx4_rxtx.h b/drivers/net/mlx4/mlx4_rxtx.h
index 2f9d3798b..4c025e3a0 100644
--- a/drivers/net/mlx4/mlx4_rxtx.h
+++ b/drivers/net/mlx4/mlx4_rxtx.h
@@ -25,6 +25,7 @@
 
 #include "mlx4.h"
 #include "mlx4_prm.h"
+#include "mlx4_mr.h"
 
 /** Rx queue counters. */
 struct mlx4_rxq_stats {
@@ -46,6 +47,7 @@ struct rxq {
 	uint16_t port_id; /**< Port ID for incoming packets. */
 	uint16_t sges_n; /**< Number of segments per packet (log2 value). */
 	uint16_t elts_n; /**< Mbuf queue size (log2 value). */
+	struct mlx4_mr_ctrl mr_ctrl; /* MR control descriptor. */
 	struct rte_mbuf *(*elts)[]; /**< Rx elements. */
 	volatile struct mlx4_wqe_data_seg (*wqes)[]; /**< HW queue entries. */
 	volatile uint32_t *rq_db; /**< RQ doorbell record. */
@@ -100,6 +102,7 @@ struct txq {
 	int elts_comp_cd; /**< Countdown for next completion. */
 	unsigned int elts_comp_cd_init; /**< Initial value for countdown. */
 	unsigned int elts_n; /**< (*elts)[] length. */
+	struct mlx4_mr_ctrl mr_ctrl; /* MR control descriptor. */
 	struct txq_elt (*elts)[]; /**< Tx elements. */
 	struct mlx4_txq_stats stats; /**< Tx queue counters. */
 	uint32_t max_inline; /**< Max inline send size. */
@@ -155,12 +158,70 @@ int mlx4_tx_queue_setup(struct rte_eth_dev *dev, uint16_t idx,
 			const struct rte_eth_txconf *conf);
 void mlx4_tx_queue_release(void *dpdk_txq);
 
-static inline uint32_t
-mlx4_txq_mp2mr(struct txq *txq, struct rte_mempool *mp)
+/* mlx4_mr.c */
+
+void mlx4_mr_flush_local_cache(struct mlx4_mr_ctrl *mr_ctrl);
+uint32_t mlx4_rx_addr2mr_bh(struct rxq *rxq, uintptr_t addr);
+uint32_t mlx4_tx_addr2mr_bh(struct txq *txq, uintptr_t addr);
+
+/**
+ * Query LKey from a packet buffer for Rx. No need to flush local caches for Rx
+ * as mempool is pre-configured and static.
+ *
+ * @param rxq
+ *   Pointer to Rx queue structure.
+ * @param addr
+ *   Address to search.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static __rte_always_inline uint32_t
+mlx4_rx_addr2mr(struct rxq *rxq, uintptr_t addr)
+{
+	struct mlx4_mr_ctrl *mr_ctrl = &rxq->mr_ctrl;
+	uint32_t lkey;
+
+	/* Linear search on MR cache array. */
+	lkey = mlx4_mr_lookup_cache(mr_ctrl->cache, &mr_ctrl->mru,
+				    MLX4_MR_CACHE_N, addr);
+	if (likely(lkey != UINT32_MAX))
+		return lkey;
+	/* Take slower bottom-half (Binary Search) on miss. */
+	return mlx4_rx_addr2mr_bh(rxq, addr);
+}
+
+#define mlx4_rx_mb2mr(rxq, mb) mlx4_rx_addr2mr(rxq, (uintptr_t)((mb)->buf_addr))
+
+/**
+ * Query LKey from a packet buffer for Tx. If not found, add the mempool.
+ *
+ * @param txq
+ *   Pointer to Tx queue structure.
+ * @param addr
+ *   Address to search.
+ *
+ * @return
+ *   Searched LKey on success, UINT32_MAX on no match.
+ */
+static __rte_always_inline uint32_t
+mlx4_tx_addr2mr(struct txq *txq, uintptr_t addr)
 {
-	(void)txq;
-	(void)mp;
-	return UINT32_MAX;
+	struct mlx4_mr_ctrl *mr_ctrl = &txq->mr_ctrl;
+	uint32_t lkey;
+
+	/* Check generation bit to see if there's any change on existing MRs. */
+	if (unlikely(*mr_ctrl->dev_gen_ptr != mr_ctrl->cur_gen))
+		mlx4_mr_flush_local_cache(mr_ctrl);
+	/* Linear search on MR cache array. */
+	lkey = mlx4_mr_lookup_cache(mr_ctrl->cache, &mr_ctrl->mru,
+				    MLX4_MR_CACHE_N, addr);
+	if (likely(lkey != UINT32_MAX))
+		return lkey;
+	/* Take slower bottom-half (binary search) on miss. */
+	return mlx4_tx_addr2mr_bh(txq, addr);
 }
 
+#define mlx4_tx_mb2mr(rxq, mb) mlx4_tx_addr2mr(rxq, (uintptr_t)((mb)->buf_addr))
+
 #endif /* MLX4_RXTX_H_ */
diff --git a/drivers/net/mlx4/mlx4_txq.c b/drivers/net/mlx4/mlx4_txq.c
index 5ea09b0b0..337ed9a1a 100644
--- a/drivers/net/mlx4/mlx4_txq.c
+++ b/drivers/net/mlx4/mlx4_txq.c
@@ -346,6 +346,13 @@ mlx4_tx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t desc,
 	/* Save first wqe pointer in the first element. */
 	(&(*txq->elts)[0])->wqe =
 		(volatile struct mlx4_wqe_ctrl_seg *)txq->msq.buf;
+	if (mlx4_mr_btree_init(&txq->mr_ctrl.cache_bh,
+			       MLX4_MR_BTREE_CACHE_N, socket)) {
+		/* rte_errno is already set. */
+		goto error;
+	}
+	/* Save pointer of global generation number to check memory event. */
+	txq->mr_ctrl.dev_gen_ptr = &priv->mr.dev_gen;
 	DEBUG("%p: adding Tx queue %p to list", (void *)dev, (void *)txq);
 	dev->data->tx_queues[idx] = txq;
 	return 0;
@@ -386,5 +393,6 @@ mlx4_tx_queue_release(void *dpdk_txq)
 		claim_zero(mlx4_glue->destroy_qp(txq->qp));
 	if (txq->cq)
 		claim_zero(mlx4_glue->destroy_cq(txq->cq));
+	mlx4_mr_btree_free(&txq->mr_ctrl.cache_bh);
 	rte_free(txq);
 }
-- 
2.11.0