1 files changed, 451 insertions, 123 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx4/mr.c b/drivers/net/ethernet/mellanox/mlx4/mr.c
index 9c188bdd7f4..2839abb878a 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mr.c
+++ b/drivers/net/ethernet/mellanox/mlx4/mr.c
@@ -33,47 +33,16 @@
  */
 
 #include <linux/errno.h>
+#include <linux/export.h>
 #include <linux/slab.h>
+#include <linux/kernel.h>
+#include <linux/vmalloc.h>
 
 #include <linux/mlx4/cmd.h>
 
 #include "mlx4.h"
 #include "icm.h"
 
-/*
- * Must be packed because mtt_seg is 64 bits but only aligned to 32 bits.
- */
-struct mlx4_mpt_entry {
-	__be32 flags;
-	__be32 qpn;
-	__be32 key;
-	__be32 pd_flags;
-	__be64 start;
-	__be64 length;
-	__be32 lkey;
-	__be32 win_cnt;
-	u8	reserved1[3];
-	u8	mtt_rep;
-	__be64 mtt_seg;
-	__be32 mtt_sz;
-	__be32 entity_size;
-	__be32 first_byte_offset;
-} __packed;
-
-#define MLX4_MPT_FLAG_SW_OWNS	    (0xfUL << 28)
-#define MLX4_MPT_FLAG_FREE	    (0x3UL << 28)
-#define MLX4_MPT_FLAG_MIO	    (1 << 17)
-#define MLX4_MPT_FLAG_BIND_ENABLE   (1 << 15)
-#define MLX4_MPT_FLAG_PHYSICAL	    (1 <<  9)
-#define MLX4_MPT_FLAG_REGION	    (1 <<  8)
-
-#define MLX4_MPT_PD_FLAG_FAST_REG   (1 << 27)
-#define MLX4_MPT_PD_FLAG_RAE	    (1 << 28)
-#define MLX4_MPT_PD_FLAG_EN_INV	    (3 << 24)
-
-#define MLX4_MPT_STATUS_SW		0xF0
-#define MLX4_MPT_STATUS_HW		0x00
-
 static u32 mlx4_buddy_alloc(struct mlx4_buddy *buddy, int order)
 {
 	int o;
@@ -137,19 +106,21 @@ static int mlx4_buddy_init(struct mlx4_buddy *buddy, int max_order)
 	buddy->max_order = max_order;
 	spin_lock_init(&buddy->lock);
 
-	buddy->bits = kzalloc((buddy->max_order + 1) * sizeof (long *),
+	buddy->bits = kcalloc(buddy->max_order + 1, sizeof (long *),
 			      GFP_KERNEL);
-	buddy->num_free = kzalloc((buddy->max_order + 1) * sizeof (int *),
+	buddy->num_free = kcalloc((buddy->max_order + 1), sizeof *buddy->num_free,
 				  GFP_KERNEL);
 	if (!buddy->bits || !buddy->num_free)
 		goto err_out;
 
 	for (i = 0; i <= buddy->max_order; ++i) {
 		s = BITS_TO_LONGS(1 << (buddy->max_order - i));
-		buddy->bits[i] = kmalloc(s * sizeof (long), GFP_KERNEL);
-		if (!buddy->bits[i])
-			goto err_out_free;
-		bitmap_zero(buddy->bits[i], 1 << (buddy->max_order - i));
+		buddy->bits[i] = kcalloc(s, sizeof (long), GFP_KERNEL | __GFP_NOWARN);
+		if (!buddy->bits[i]) {
+			buddy->bits[i] = vzalloc(s * sizeof(long));
+			if (!buddy->bits[i])
+				goto err_out_free;
+		}
 	}
 
 	set_bit(0, buddy->bits[buddy->max_order]);
@@ -159,7 +130,10 @@ static int mlx4_buddy_init(struct mlx4_buddy *buddy, int max_order)
 
 err_out_free:
 	for (i = 0; i <= buddy->max_order; ++i)
-		kfree(buddy->bits[i]);
+		if (buddy->bits[i] && is_vmalloc_addr(buddy->bits[i]))
+			vfree(buddy->bits[i]);
+		else
+			kfree(buddy->bits[i]);
 
 err_out:
 	kfree(buddy->bits);
@@ -173,28 +147,57 @@ static void mlx4_buddy_cleanup(struct mlx4_buddy *buddy)
 	int i;
 
 	for (i = 0; i <= buddy->max_order; ++i)
-		kfree(buddy->bits[i]);
+		if (is_vmalloc_addr(buddy->bits[i]))
+			vfree(buddy->bits[i]);
+		else
+			kfree(buddy->bits[i]);
 
 	kfree(buddy->bits);
 	kfree(buddy->num_free);
 }
 
-static u32 mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
+u32 __mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
 {
 	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
 	u32 seg;
+	int seg_order;
+	u32 offset;
 
-	seg = mlx4_buddy_alloc(&mr_table->mtt_buddy, order);
+	seg_order = max_t(int, order - log_mtts_per_seg, 0);
+
+	seg = mlx4_buddy_alloc(&mr_table->mtt_buddy, seg_order);
 	if (seg == -1)
 		return -1;
 
-	if (mlx4_table_get_range(dev, &mr_table->mtt_table, seg,
-				 seg + (1 << order) - 1)) {
-		mlx4_buddy_free(&mr_table->mtt_buddy, seg, order);
+	offset = seg * (1 << log_mtts_per_seg);
+
+	if (mlx4_table_get_range(dev, &mr_table->mtt_table, offset,
+				 offset + (1 << order) - 1)) {
+		mlx4_buddy_free(&mr_table->mtt_buddy, seg, seg_order);
 		return -1;
 	}
 
-	return seg;
+	return offset;
+}
+
+static u32 mlx4_alloc_mtt_range(struct mlx4_dev *dev, int order)
+{
+	u64 in_param = 0;
+	u64 out_param;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, order);
+		err = mlx4_cmd_imm(dev, in_param, &out_param, RES_MTT,
+						       RES_OP_RESERVE_AND_MAP,
+						       MLX4_CMD_ALLOC_RES,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_WRAPPED);
+		if (err)
+			return -1;
+		return get_param_l(&out_param);
+	}
+	return __mlx4_alloc_mtt_range(dev, order);
 }
 
 int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,
@@ -209,33 +212,63 @@ int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,
 	} else
 		mtt->page_shift = page_shift;
 
-	for (mtt->order = 0, i = dev->caps.mtts_per_seg; i < npages; i <<= 1)
+	for (mtt->order = 0, i = 1; i < npages; i <<= 1)
 		++mtt->order;
 
-	mtt->first_seg = mlx4_alloc_mtt_range(dev, mtt->order);
-	if (mtt->first_seg == -1)
+	mtt->offset = mlx4_alloc_mtt_range(dev, mtt->order);
+	if (mtt->offset == -1)
 		return -ENOMEM;
 
 	return 0;
 }
 EXPORT_SYMBOL_GPL(mlx4_mtt_init);
 
-void mlx4_mtt_cleanup(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+void __mlx4_free_mtt_range(struct mlx4_dev *dev, u32 offset, int order)
 {
+	u32 first_seg;
+	int seg_order;
 	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
 
+	seg_order = max_t(int, order - log_mtts_per_seg, 0);
+	first_seg = offset / (1 << log_mtts_per_seg);
+
+	mlx4_buddy_free(&mr_table->mtt_buddy, first_seg, seg_order);
+	mlx4_table_put_range(dev, &mr_table->mtt_table, offset,
+			     offset + (1 << order) - 1);
+}
+
+static void mlx4_free_mtt_range(struct mlx4_dev *dev, u32 offset, int order)
+{
+	u64 in_param = 0;
+	int err;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, offset);
+		set_param_h(&in_param, order);
+		err = mlx4_cmd(dev, in_param, RES_MTT, RES_OP_RESERVE_AND_MAP,
+						       MLX4_CMD_FREE_RES,
+						       MLX4_CMD_TIME_CLASS_A,
+						       MLX4_CMD_WRAPPED);
+		if (err)
+			mlx4_warn(dev, "Failed to free mtt range at:%d order:%d\n",
+				  offset, order);
+		return;
+	}
+	 __mlx4_free_mtt_range(dev, offset, order);
+}
+
+void mlx4_mtt_cleanup(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
+{
 	if (mtt->order < 0)
 		return;
 
-	mlx4_buddy_free(&mr_table->mtt_buddy, mtt->first_seg, mtt->order);
-	mlx4_table_put_range(dev, &mr_table->mtt_table, mtt->first_seg,
-			     mtt->first_seg + (1 << mtt->order) - 1);
+	mlx4_free_mtt_range(dev, mtt->offset, mtt->order);
 }
 EXPORT_SYMBOL_GPL(mlx4_mtt_cleanup);
 
 u64 mlx4_mtt_addr(struct mlx4_dev *dev, struct mlx4_mtt *mtt)
 {
-	return (u64) mtt->first_seg * dev->caps.mtt_entry_sz;
+	return (u64) mtt->offset * dev->caps.mtt_entry_sz;
 }
 EXPORT_SYMBOL_GPL(mlx4_mtt_addr);
 
@@ -252,69 +285,191 @@ static u32 key_to_hw_index(u32 key)
 static int mlx4_SW2HW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
 			  int mpt_index)
 {
-	return mlx4_cmd(dev, mailbox->dma, mpt_index, 0, MLX4_CMD_SW2HW_MPT,
-			MLX4_CMD_TIME_CLASS_B);
+	return mlx4_cmd(dev, mailbox->dma, mpt_index,
+			0, MLX4_CMD_SW2HW_MPT, MLX4_CMD_TIME_CLASS_B,
+			MLX4_CMD_WRAPPED);
 }
 
 static int mlx4_HW2SW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
 			  int mpt_index)
 {
 	return mlx4_cmd_box(dev, 0, mailbox ? mailbox->dma : 0, mpt_index,
-			    !mailbox, MLX4_CMD_HW2SW_MPT, MLX4_CMD_TIME_CLASS_B);
+			    !mailbox, MLX4_CMD_HW2SW_MPT,
+			    MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+}
+
+static int mlx4_mr_alloc_reserved(struct mlx4_dev *dev, u32 mridx, u32 pd,
+			   u64 iova, u64 size, u32 access, int npages,
+			   int page_shift, struct mlx4_mr *mr)
+{
+	mr->iova       = iova;
+	mr->size       = size;
+	mr->pd	       = pd;
+	mr->access     = access;
+	mr->enabled    = MLX4_MPT_DISABLED;
+	mr->key	       = hw_index_to_key(mridx);
+
+	return mlx4_mtt_init(dev, npages, page_shift, &mr->mtt);
+}
+
+static int mlx4_WRITE_MTT(struct mlx4_dev *dev,
+			  struct mlx4_cmd_mailbox *mailbox,
+			  int num_entries)
+{
+	return mlx4_cmd(dev, mailbox->dma, num_entries, 0, MLX4_CMD_WRITE_MTT,
+			MLX4_CMD_TIME_CLASS_A,  MLX4_CMD_WRAPPED);
+}
+
+int __mlx4_mpt_reserve(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	return mlx4_bitmap_alloc(&priv->mr_table.mpt_bitmap);
+}
+
+static int mlx4_mpt_reserve(struct mlx4_dev *dev)
+{
+	u64 out_param;
+
+	if (mlx4_is_mfunc(dev)) {
+		if (mlx4_cmd_imm(dev, 0, &out_param, RES_MPT, RES_OP_RESERVE,
+				   MLX4_CMD_ALLOC_RES,
+				   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED))
+			return -1;
+		return get_param_l(&out_param);
+	}
+	return  __mlx4_mpt_reserve(dev);
+}
+
+void __mlx4_mpt_release(struct mlx4_dev *dev, u32 index)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+
+	mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, index, MLX4_NO_RR);
+}
+
+static void mlx4_mpt_release(struct mlx4_dev *dev, u32 index)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, index);
+		if (mlx4_cmd(dev, in_param, RES_MPT, RES_OP_RESERVE,
+			       MLX4_CMD_FREE_RES,
+			       MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed to release mr index:%d\n",
+				  index);
+		return;
+	}
+	__mlx4_mpt_release(dev, index);
+}
+
+int __mlx4_mpt_alloc_icm(struct mlx4_dev *dev, u32 index, gfp_t gfp)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	return mlx4_table_get(dev, &mr_table->dmpt_table, index, gfp);
+}
+
+static int mlx4_mpt_alloc_icm(struct mlx4_dev *dev, u32 index, gfp_t gfp)
+{
+	u64 param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&param, index);
+		return mlx4_cmd_imm(dev, param, &param, RES_MPT, RES_OP_MAP_ICM,
+							MLX4_CMD_ALLOC_RES,
+							MLX4_CMD_TIME_CLASS_A,
+							MLX4_CMD_WRAPPED);
+	}
+	return __mlx4_mpt_alloc_icm(dev, index, gfp);
+}
+
+void __mlx4_mpt_free_icm(struct mlx4_dev *dev, u32 index)
+{
+	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+
+	mlx4_table_put(dev, &mr_table->dmpt_table, index);
+}
+
+static void mlx4_mpt_free_icm(struct mlx4_dev *dev, u32 index)
+{
+	u64 in_param = 0;
+
+	if (mlx4_is_mfunc(dev)) {
+		set_param_l(&in_param, index);
+		if (mlx4_cmd(dev, in_param, RES_MPT, RES_OP_MAP_ICM,
+			     MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+			     MLX4_CMD_WRAPPED))
+			mlx4_warn(dev, "Failed to free icm of mr index:%d\n",
+				  index);
+		return;
+	}
+	return __mlx4_mpt_free_icm(dev, index);
 }
 
 int mlx4_mr_alloc(struct mlx4_dev *dev, u32 pd, u64 iova, u64 size, u32 access,
 		  int npages, int page_shift, struct mlx4_mr *mr)
 {
-	struct mlx4_priv *priv = mlx4_priv(dev);
 	u32 index;
 	int err;
 
-	index = mlx4_bitmap_alloc(&priv->mr_table.mpt_bitmap);
+	index = mlx4_mpt_reserve(dev);
 	if (index == -1)
 		return -ENOMEM;
 
-	mr->iova       = iova;
-	mr->size       = size;
-	mr->pd	       = pd;
-	mr->access     = access;
-	mr->enabled    = 0;
-	mr->key	       = hw_index_to_key(index);
-
-	err = mlx4_mtt_init(dev, npages, page_shift, &mr->mtt);
+	err = mlx4_mr_alloc_reserved(dev, index, pd, iova, size,
+				     access, npages, page_shift, mr);
 	if (err)
-		mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, index);
+		mlx4_mpt_release(dev, index);
 
 	return err;
 }
 EXPORT_SYMBOL_GPL(mlx4_mr_alloc);
 
-void mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr)
+static int mlx4_mr_free_reserved(struct mlx4_dev *dev, struct mlx4_mr *mr)
 {
-	struct mlx4_priv *priv = mlx4_priv(dev);
 	int err;
 
-	if (mr->enabled) {
+	if (mr->enabled == MLX4_MPT_EN_HW) {
 		err = mlx4_HW2SW_MPT(dev, NULL,
 				     key_to_hw_index(mr->key) &
 				     (dev->caps.num_mpts - 1));
-		if (err)
-			mlx4_warn(dev, "HW2SW_MPT failed (%d)\n", err);
-	}
+		if (err) {
+			mlx4_warn(dev, "HW2SW_MPT failed (%d), MR has MWs bound to it\n",
+				  err);
+			return err;
+		}
 
+		mr->enabled = MLX4_MPT_EN_SW;
+	}
 	mlx4_mtt_cleanup(dev, &mr->mtt);
-	mlx4_bitmap_free(&priv->mr_table.mpt_bitmap, key_to_hw_index(mr->key));
+
+	return 0;
+}
+
+int mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr)
+{
+	int ret;
+
+	ret = mlx4_mr_free_reserved(dev, mr);
+	if (ret)
+		return ret;
+	if (mr->enabled)
+		mlx4_mpt_free_icm(dev, key_to_hw_index(mr->key));
+	mlx4_mpt_release(dev, key_to_hw_index(mr->key));
+
+	return 0;
 }
 EXPORT_SYMBOL_GPL(mlx4_mr_free);
 
 int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
 {
-	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
 	struct mlx4_cmd_mailbox *mailbox;
 	struct mlx4_mpt_entry *mpt_entry;
 	int err;
 
-	err = mlx4_table_get(dev, &mr_table->dmpt_table, key_to_hw_index(mr->key));
+	err = mlx4_mpt_alloc_icm(dev, key_to_hw_index(mr->key), GFP_KERNEL);
 	if (err)
 		return err;
 
@@ -324,9 +479,6 @@ int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
 		goto err_table;
 	}
 	mpt_entry = mailbox->buf;
-
-	memset(mpt_entry, 0, sizeof *mpt_entry);
-
 	mpt_entry->flags = cpu_to_be32(MLX4_MPT_FLAG_MIO	 |
 				       MLX4_MPT_FLAG_REGION	 |
 				       mr->access);
@@ -339,9 +491,10 @@ int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
 
 	if (mr->mtt.order < 0) {
 		mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL);
-		mpt_entry->mtt_seg = 0;
+		mpt_entry->mtt_addr = 0;
 	} else {
-		mpt_entry->mtt_seg = cpu_to_be64(mlx4_mtt_addr(dev, &mr->mtt));
+		mpt_entry->mtt_addr = cpu_to_be64(mlx4_mtt_addr(dev,
+						  &mr->mtt));
 	}
 
 	if (mr->mtt.order >= 0 && mr->mtt.page_shift == 0) {
@@ -349,8 +502,7 @@ int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
 		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_FREE);
 		mpt_entry->pd_flags |= cpu_to_be32(MLX4_MPT_PD_FLAG_FAST_REG |
 						   MLX4_MPT_PD_FLAG_RAE);
-		mpt_entry->mtt_sz    = cpu_to_be32((1 << mr->mtt.order) *
-						   dev->caps.mtts_per_seg);
+		mpt_entry->mtt_sz    = cpu_to_be32(1 << mr->mtt.order);
 	} else {
 		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_SW_OWNS);
 	}
@@ -361,8 +513,7 @@ int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr)
 		mlx4_warn(dev, "SW2HW_MPT failed (%d)\n", err);
 		goto err_cmd;
 	}
-
-	mr->enabled = 1;
+	mr->enabled = MLX4_MPT_EN_HW;
 
 	mlx4_free_cmd_mailbox(dev, mailbox);
 
@@ -372,7 +523,7 @@ err_cmd:
 	mlx4_free_cmd_mailbox(dev, mailbox);
 
 err_table:
-	mlx4_table_put(dev, &mr_table->dmpt_table, key_to_hw_index(mr->key));
+	mlx4_mpt_free_icm(dev, key_to_hw_index(mr->key));
 	return err;
 }
 EXPORT_SYMBOL_GPL(mlx4_mr_enable);
@@ -384,18 +535,10 @@ static int mlx4_write_mtt_chunk(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
 	__be64 *mtts;
 	dma_addr_t dma_handle;
 	int i;
-	int s = start_index * sizeof (u64);
-
-	/* All MTTs must fit in the same page */
-	if (start_index / (PAGE_SIZE / sizeof (u64)) !=
-	    (start_index + npages - 1) / (PAGE_SIZE / sizeof (u64)))
-		return -EINVAL;
 
-	if (start_index & (dev->caps.mtts_per_seg - 1))
-		return -EINVAL;
+	mtts = mlx4_table_find(&priv->mr_table.mtt_table, mtt->offset +
+			       start_index, &dma_handle);
 
-	mtts = mlx4_table_find(&priv->mr_table.mtt_table, mtt->first_seg +
-				s / dev->caps.mtt_entry_sz, &dma_handle);
 	if (!mtts)
 		return -ENOMEM;
 
@@ -411,38 +554,87 @@ static int mlx4_write_mtt_chunk(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
 	return 0;
 }
 
-int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
-		   int start_index, int npages, u64 *page_list)
+int __mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		     int start_index, int npages, u64 *page_list)
 {
+	int err = 0;
 	int chunk;
-	int err;
+	int mtts_per_page;
+	int max_mtts_first_page;
 
-	if (mtt->order < 0)
-		return -EINVAL;
+	/* compute how may mtts fit in the first page */
+	mtts_per_page = PAGE_SIZE / sizeof(u64);
+	max_mtts_first_page = mtts_per_page - (mtt->offset + start_index)
+			      % mtts_per_page;
+
+	chunk = min_t(int, max_mtts_first_page, npages);
 
 	while (npages > 0) {
-		chunk = min_t(int, PAGE_SIZE / sizeof(u64), npages);
 		err = mlx4_write_mtt_chunk(dev, mtt, start_index, chunk, page_list);
 		if (err)
 			return err;
-
 		npages      -= chunk;
 		start_index += chunk;
 		page_list   += chunk;
+
+		chunk = min_t(int, mtts_per_page, npages);
 	}
+	return err;
+}
 
-	return 0;
+int mlx4_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
+		   int start_index, int npages, u64 *page_list)
+{
+	struct mlx4_cmd_mailbox *mailbox = NULL;
+	__be64 *inbox = NULL;
+	int chunk;
+	int err = 0;
+	int i;
+
+	if (mtt->order < 0)
+		return -EINVAL;
+
+	if (mlx4_is_mfunc(dev)) {
+		mailbox = mlx4_alloc_cmd_mailbox(dev);
+		if (IS_ERR(mailbox))
+			return PTR_ERR(mailbox);
+		inbox = mailbox->buf;
+
+		while (npages > 0) {
+			chunk = min_t(int, MLX4_MAILBOX_SIZE / sizeof(u64) - 2,
+				      npages);
+			inbox[0] = cpu_to_be64(mtt->offset + start_index);
+			inbox[1] = 0;
+			for (i = 0; i < chunk; ++i)
+				inbox[i + 2] = cpu_to_be64(page_list[i] |
+					       MLX4_MTT_FLAG_PRESENT);
+			err = mlx4_WRITE_MTT(dev, mailbox, chunk);
+			if (err) {
+				mlx4_free_cmd_mailbox(dev, mailbox);
+				return err;
+			}
+
+			npages      -= chunk;
+			start_index += chunk;
+			page_list   += chunk;
+		}
+		mlx4_free_cmd_mailbox(dev, mailbox);
+		return err;
+	}
+
+	return __mlx4_write_mtt(dev, mtt, start_index, npages, page_list);
 }
 EXPORT_SYMBOL_GPL(mlx4_write_mtt);
 
 int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
-		       struct mlx4_buf *buf)
+		       struct mlx4_buf *buf, gfp_t gfp)
 {
 	u64 *page_list;
 	int err;
 	int i;
 
-	page_list = kmalloc(buf->npages * sizeof *page_list, GFP_KERNEL);
+	page_list = kmalloc(buf->npages * sizeof *page_list,
+			    gfp);
 	if (!page_list)
 		return -ENOMEM;
 
@@ -459,24 +651,130 @@ int mlx4_buf_write_mtt(struct mlx4_dev *dev, struct mlx4_mtt *mtt,
 }
 EXPORT_SYMBOL_GPL(mlx4_buf_write_mtt);
 
+int mlx4_mw_alloc(struct mlx4_dev *dev, u32 pd, enum mlx4_mw_type type,
+		  struct mlx4_mw *mw)
+{
+	u32 index;
+
+	if ((type == MLX4_MW_TYPE_1 &&
+	     !(dev->caps.flags & MLX4_DEV_CAP_FLAG_MEM_WINDOW)) ||
+	     (type == MLX4_MW_TYPE_2 &&
+	     !(dev->caps.bmme_flags & MLX4_BMME_FLAG_TYPE_2_WIN)))
+		return -ENOTSUPP;
+
+	index = mlx4_mpt_reserve(dev);
+	if (index == -1)
+		return -ENOMEM;
+
+	mw->key	    = hw_index_to_key(index);
+	mw->pd      = pd;
+	mw->type    = type;
+	mw->enabled = MLX4_MPT_DISABLED;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_alloc);
+
+int mlx4_mw_enable(struct mlx4_dev *dev, struct mlx4_mw *mw)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_mpt_entry *mpt_entry;
+	int err;
+
+	err = mlx4_mpt_alloc_icm(dev, key_to_hw_index(mw->key), GFP_KERNEL);
+	if (err)
+		return err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		goto err_table;
+	}
+	mpt_entry = mailbox->buf;
+
+	/* Note that the MLX4_MPT_FLAG_REGION bit in mpt_entry->flags is turned
+	 * off, thus creating a memory window and not a memory region.
+	 */
+	mpt_entry->key	       = cpu_to_be32(key_to_hw_index(mw->key));
+	mpt_entry->pd_flags    = cpu_to_be32(mw->pd);
+	if (mw->type == MLX4_MW_TYPE_2) {
+		mpt_entry->flags    |= cpu_to_be32(MLX4_MPT_FLAG_FREE);
+		mpt_entry->qpn       = cpu_to_be32(MLX4_MPT_QP_FLAG_BOUND_QP);
+		mpt_entry->pd_flags |= cpu_to_be32(MLX4_MPT_PD_FLAG_EN_INV);
+	}
+
+	err = mlx4_SW2HW_MPT(dev, mailbox,
+			     key_to_hw_index(mw->key) &
+			     (dev->caps.num_mpts - 1));
+	if (err) {
+		mlx4_warn(dev, "SW2HW_MPT failed (%d)\n", err);
+		goto err_cmd;
+	}
+	mw->enabled = MLX4_MPT_EN_HW;
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+	return 0;
+
+err_cmd:
+	mlx4_free_cmd_mailbox(dev, mailbox);
+
+err_table:
+	mlx4_mpt_free_icm(dev, key_to_hw_index(mw->key));
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_enable);
+
+void mlx4_mw_free(struct mlx4_dev *dev, struct mlx4_mw *mw)
+{
+	int err;
+
+	if (mw->enabled == MLX4_MPT_EN_HW) {
+		err = mlx4_HW2SW_MPT(dev, NULL,
+				     key_to_hw_index(mw->key) &
+				     (dev->caps.num_mpts - 1));
+		if (err)
+			mlx4_warn(dev, "xxx HW2SW_MPT failed (%d)\n", err);
+
+		mw->enabled = MLX4_MPT_EN_SW;
+	}
+	if (mw->enabled)
+		mlx4_mpt_free_icm(dev, key_to_hw_index(mw->key));
+	mlx4_mpt_release(dev, key_to_hw_index(mw->key));
+}
+EXPORT_SYMBOL_GPL(mlx4_mw_free);
+
 int mlx4_init_mr_table(struct mlx4_dev *dev)
 {
-	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mr_table *mr_table = &priv->mr_table;
 	int err;
 
+	/* Nothing to do for slaves - all MR handling is forwarded
+	* to the master */
+	if (mlx4_is_slave(dev))
+		return 0;
+
+	if (!is_power_of_2(dev->caps.num_mpts))
+		return -EINVAL;
+
 	err = mlx4_bitmap_init(&mr_table->mpt_bitmap, dev->caps.num_mpts,
 			       ~0, dev->caps.reserved_mrws, 0);
 	if (err)
 		return err;
 
 	err = mlx4_buddy_init(&mr_table->mtt_buddy,
-			      ilog2(dev->caps.num_mtt_segs));
+			      ilog2((u32)dev->caps.num_mtts /
+			      (1 << log_mtts_per_seg)));
 	if (err)
 		goto err_buddy;
 
 	if (dev->caps.reserved_mtts) {
-		if (mlx4_alloc_mtt_range(dev, fls(dev->caps.reserved_mtts - 1)) == -1) {
-			mlx4_warn(dev, "MTT table of order %d is too small.\n",
+		priv->reserved_mtts =
+			mlx4_alloc_mtt_range(dev,
+					     fls(dev->caps.reserved_mtts - 1));
+		if (priv->reserved_mtts < 0) {
+			mlx4_warn(dev, "MTT table of order %u is too small\n",
 				  mr_table->mtt_buddy.max_order);
 			err = -ENOMEM;
 			goto err_reserve_mtts;
@@ -496,8 +794,14 @@ err_buddy:
 
 void mlx4_cleanup_mr_table(struct mlx4_dev *dev)
 {
-	struct mlx4_mr_table *mr_table = &mlx4_priv(dev)->mr_table;
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_mr_table *mr_table = &priv->mr_table;
 
+	if (mlx4_is_slave(dev))
+		return;
+	if (priv->reserved_mtts >= 0)
+		mlx4_free_mtt_range(dev, priv->reserved_mtts,
+				    fls(dev->caps.reserved_mtts - 1));
 	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
 	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
 }
@@ -580,9 +884,11 @@ int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages,
 		   int max_maps, u8 page_shift, struct mlx4_fmr *fmr)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
-	u64 mtt_seg;
 	int err = -ENOMEM;
 
+	if (max_maps > dev->caps.max_fmr_maps)
+		return -EINVAL;
+
 	if (page_shift < (ffs(dev->caps.page_size_cap) - 1) || page_shift >= 32)
 		return -EINVAL;
 
@@ -600,11 +906,10 @@ int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages,
 	if (err)
 		return err;
 
-	mtt_seg = fmr->mr.mtt.first_seg * dev->caps.mtt_entry_sz;
-
 	fmr->mtts = mlx4_table_find(&priv->mr_table.mtt_table,
-				    fmr->mr.mtt.first_seg,
+				    fmr->mr.mtt.offset,
 				    &fmr->dma_handle);
+
 	if (!fmr->mtts) {
 		err = -ENOMEM;
 		goto err_free;
@@ -613,7 +918,7 @@ int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages,
 	return 0;
 
 err_free:
-	mlx4_mr_free(dev, &fmr->mr);
+	(void) mlx4_mr_free(dev, &fmr->mr);
 	return err;
 }
 EXPORT_SYMBOL_GPL(mlx4_fmr_alloc);
@@ -639,22 +944,44 @@ EXPORT_SYMBOL_GPL(mlx4_fmr_enable);
 void mlx4_fmr_unmap(struct mlx4_dev *dev, struct mlx4_fmr *fmr,
 		    u32 *lkey, u32 *rkey)
 {
+	struct mlx4_cmd_mailbox *mailbox;
+	int err;
+
 	if (!fmr->maps)
 		return;
 
 	fmr->maps = 0;
 
-	*(u8 *) fmr->mpt = MLX4_MPT_STATUS_SW;
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox)) {
+		err = PTR_ERR(mailbox);
+		pr_warn("mlx4_ib: mlx4_alloc_cmd_mailbox failed (%d)\n", err);
+		return;
+	}
+
+	err = mlx4_HW2SW_MPT(dev, NULL,
+			     key_to_hw_index(fmr->mr.key) &
+			     (dev->caps.num_mpts - 1));
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	if (err) {
+		pr_warn("mlx4_ib: mlx4_HW2SW_MPT failed (%d)\n", err);
+		return;
+	}
+	fmr->mr.enabled = MLX4_MPT_EN_SW;
 }
 EXPORT_SYMBOL_GPL(mlx4_fmr_unmap);
 
 int mlx4_fmr_free(struct mlx4_dev *dev, struct mlx4_fmr *fmr)
 {
+	int ret;
+
 	if (fmr->maps)
 		return -EBUSY;
 
-	fmr->mr.enabled = 0;
-	mlx4_mr_free(dev, &fmr->mr);
+	ret = mlx4_mr_free(dev, &fmr->mr);
+	if (ret)
+		return ret;
+	fmr->mr.enabled = MLX4_MPT_DISABLED;
 
 	return 0;
 }
@@ -662,6 +989,7 @@ EXPORT_SYMBOL_GPL(mlx4_fmr_free);
 
 int mlx4_SYNC_TPT(struct mlx4_dev *dev)
 {
-	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_SYNC_TPT, 1000);
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_SYNC_TPT, 1000,
+			MLX4_CMD_NATIVE);
 }
 EXPORT_SYMBOL_GPL(mlx4_SYNC_TPT);