diff mbox

[for-next,1/8] mlx5: Mellanox Connect-IB driver part 1/8

Message ID 1371384149-24558-2-git-send-email-eli@dev.mellanox.co.il (mailing list archive)
State Superseded, archived
Delegated to: Roland Dreier
Headers show

Commit Message

Eli Cohen June 16, 2013, 12:02 p.m. UTC
From: Eli Cohen <eli@mellanox.com>

Signed-off-by: Eli Cohen <eli@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/alloc.c   |  244 ++++
 drivers/net/ethernet/mellanox/mlx5/core/cmd.c     | 1497 +++++++++++++++++++++
 drivers/net/ethernet/mellanox/mlx5/core/cq.c      |  226 ++++
 drivers/net/ethernet/mellanox/mlx5/core/debugfs.c |  600 +++++++++
 4 files changed, 2567 insertions(+)
 create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/alloc.c
 create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/cmd.c
 create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/cq.c
 create mode 100644 drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
diff mbox

Patch

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/alloc.c b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
new file mode 100644
index 0000000..047cb03
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/alloc.c
@@ -0,0 +1,244 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies 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/errno.h>
+#include <linux/slab.h>
+#include <linux/mm.h>
+#include <linux/export.h>
+#include <linux/bitmap.h>
+#include <linux/dma-mapping.h>
+#include <linux/vmalloc.h>
+#include <linux/mlx5/driver.h>
+
+#include "mlx5_core.h"
+
+MLX5_MOD_DBG_MASK(MLX5_MOD_ALLOC);
+
+/* Handling for queue buffers -- we allocate a bunch of memory and
+ * register it in a memory region at HCA virtual address 0.  If the
+ * requested size is > max_direct, we split the allocation into
+ * multiple pages, so we don't require too much contiguous memory.
+ */
+
+int mlx5_buf_alloc(struct mlx5_core_dev *dev, int size, int max_direct,
+		   struct mlx5_buf *buf)
+{
+	dma_addr_t t;
+
+	buf->size = size;
+	if (size <= max_direct) {
+		buf->nbufs        = 1;
+		buf->npages       = 1;
+		buf->page_shift   = get_order(size) + PAGE_SHIFT;
+		buf->direct.buf   = dma_alloc_coherent(&dev->pdev->dev,
+						       size, &t, GFP_KERNEL);
+		if (!buf->direct.buf)
+			return -ENOMEM;
+
+		buf->direct.map = t;
+
+		while (t & ((1 << buf->page_shift) - 1)) {
+			--buf->page_shift;
+			buf->npages *= 2;
+		}
+
+		memset(buf->direct.buf, 0, size);
+	} else {
+		int i;
+
+		buf->direct.buf  = NULL;
+		buf->nbufs       = (size + PAGE_SIZE - 1) / PAGE_SIZE;
+		buf->npages      = buf->nbufs;
+		buf->page_shift  = PAGE_SHIFT;
+		buf->page_list   = kcalloc(buf->nbufs, sizeof(*buf->page_list),
+					   GFP_KERNEL);
+		if (!buf->page_list)
+			return -ENOMEM;
+
+		for (i = 0; i < buf->nbufs; ++i) {
+			buf->page_list[i].buf =
+				dma_alloc_coherent(&dev->pdev->dev, PAGE_SIZE,
+						   &t, GFP_KERNEL);
+			if (!buf->page_list[i].buf)
+				goto err_free;
+
+			buf->page_list[i].map = t;
+
+			memset(buf->page_list[i].buf, 0, PAGE_SIZE);
+		}
+
+		if (BITS_PER_LONG == 64) {
+			struct page **pages;
+			pages = kmalloc(sizeof(*pages) * buf->nbufs, GFP_KERNEL);
+			if (!pages)
+				goto err_free;
+			for (i = 0; i < buf->nbufs; ++i)
+				pages[i] = virt_to_page(buf->page_list[i].buf);
+			buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP, PAGE_KERNEL);
+			kfree(pages);
+			if (!buf->direct.buf)
+				goto err_free;
+		}
+	}
+
+	return 0;
+
+err_free:
+	mlx5_buf_free(dev, buf);
+
+	return -ENOMEM;
+}
+EXPORT_SYMBOL_GPL(mlx5_buf_alloc);
+
+void mlx5_buf_free(struct mlx5_core_dev *dev, struct mlx5_buf *buf)
+{
+	int i;
+
+	if (buf->nbufs == 1)
+		dma_free_coherent(&dev->pdev->dev, buf->size, buf->direct.buf,
+				  buf->direct.map);
+	else {
+		if (BITS_PER_LONG == 64 && buf->direct.buf)
+			vunmap(buf->direct.buf);
+
+		for (i = 0; i < buf->nbufs; ++i)
+			if (buf->page_list[i].buf)
+				dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
+						  buf->page_list[i].buf,
+						  buf->page_list[i].map);
+		kfree(buf->page_list);
+	}
+}
+EXPORT_SYMBOL_GPL(mlx5_buf_free);
+
+static struct mlx5_db_pgdir *mlx5_alloc_db_pgdir(struct device *dma_device)
+{
+	struct mlx5_db_pgdir *pgdir;
+
+	pgdir = kzalloc(sizeof(*pgdir), GFP_KERNEL);
+	if (!pgdir)
+		return NULL;
+
+	bitmap_fill(pgdir->bitmap, MLX5_DB_PER_PAGE);
+	pgdir->db_page = dma_alloc_coherent(dma_device, PAGE_SIZE,
+					    &pgdir->db_dma, GFP_KERNEL);
+	if (!pgdir->db_page) {
+		kfree(pgdir);
+		return NULL;
+	}
+
+	return pgdir;
+}
+
+static int mlx5_alloc_db_from_pgdir(struct mlx5_db_pgdir *pgdir,
+				    struct mlx5_db *db)
+{
+	int i;
+	int offset;
+
+	i = find_first_bit(pgdir->bitmap, MLX5_DB_PER_PAGE);
+	if (i >= MLX5_DB_PER_PAGE)
+		return -ENOMEM;
+
+	clear_bit(i, pgdir->bitmap);
+
+	db->u.pgdir = pgdir;
+	db->index   = i;
+	offset = db->index * L1_CACHE_BYTES;
+	db->db      = pgdir->db_page + offset / sizeof(*pgdir->db_page);
+	db->dma     = pgdir->db_dma  + offset;
+
+	return 0;
+}
+
+int mlx5_db_alloc(struct mlx5_core_dev *dev, struct mlx5_db *db)
+{
+	struct mlx5_db_pgdir *pgdir;
+	int ret = 0;
+
+	mutex_lock(&dev->priv.pgdir_mutex);
+
+	list_for_each_entry(pgdir, &dev->priv.pgdir_list, list)
+		if (!mlx5_alloc_db_from_pgdir(pgdir, db))
+			goto out;
+
+	pgdir = mlx5_alloc_db_pgdir(&(dev->pdev->dev));
+	if (!pgdir) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	list_add(&pgdir->list, &dev->priv.pgdir_list);
+
+	/* This should never fail -- we just allocated an empty page: */
+	WARN_ON(mlx5_alloc_db_from_pgdir(pgdir, db));
+
+out:
+	mutex_unlock(&dev->priv.pgdir_mutex);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(mlx5_db_alloc);
+
+void mlx5_db_free(struct mlx5_core_dev *dev, struct mlx5_db *db)
+{
+	mutex_lock(&dev->priv.pgdir_mutex);
+
+	set_bit(db->index, db->u.pgdir->bitmap);
+
+	if (bitmap_full(db->u.pgdir->bitmap, MLX5_DB_PER_PAGE)) {
+		dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE,
+				  db->u.pgdir->db_page, db->u.pgdir->db_dma);
+		list_del(&db->u.pgdir->list);
+		kfree(db->u.pgdir);
+	}
+
+	mutex_unlock(&dev->priv.pgdir_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx5_db_free);
+
+
+void mlx5_fill_page_array(struct mlx5_buf *buf, __be64 *pas)
+{
+	int i;
+	u64 addr;
+
+	for (i = 0; i < buf->npages; ++i) {
+		if (buf->nbufs == 1)
+			addr = buf->direct.map + (i << buf->page_shift);
+		else
+			addr = buf->page_list[i].map;
+
+		pas[i] = cpu_to_be64(addr);
+	}
+}
+EXPORT_SYMBOL_GPL(mlx5_fill_page_array);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
new file mode 100644
index 0000000..33d00b7
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
@@ -0,0 +1,1497 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies 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 <asm-generic/kmap_types.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/pci.h>
+#include <linux/dma-mapping.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/random.h>
+#include <linux/io-mapping.h>
+#include <linux/mlx5/driver.h>
+#include <linux/debugfs.h>
+
+#include "mlx5_core.h"
+
+MLX5_MOD_DBG_MASK(MLX5_MOD_CMDIF);
+
+enum {
+	CMD_IF_REV = 3,
+};
+
+enum {
+	CMD_MODE_POLLING,
+	CMD_MODE_EVENTS
+};
+
+enum {
+	NUM_LONG_LISTS	  = 2,
+	NUM_MED_LISTS	  = 64,
+	LONG_LIST_SIZE	  = (2ULL * 1024 * 1024 * 1024 / PAGE_SIZE) * 8 + 16 +
+				MLX5_CMD_DATA_BLOCK_SIZE,
+	MED_LIST_SIZE	  = 16 + MLX5_CMD_DATA_BLOCK_SIZE,
+};
+
+static int dbg_open(struct inode *inode, struct file *file)
+{
+	file->private_data = inode->i_private;
+
+	return 0;
+}
+
+static struct mlx5_cmd_work_ent *alloc_cmd(struct mlx5_cmd *cmd,
+					   struct mlx5_cmd_msg *in,
+					   struct mlx5_cmd_msg *out,
+					   mlx5_cmd_cbk_t cbk,
+					   void *context, int page_queue)
+{
+	struct mlx5_cmd_work_ent *ent;
+	gfp_t alloc_flags = cbk ? GFP_ATOMIC : GFP_KERNEL;
+
+	ent = kzalloc(sizeof(*ent), alloc_flags);
+	if (!ent)
+		return ERR_PTR(-ENOMEM);
+
+	ent->in		= in;
+	ent->out	= out;
+	ent->callback	= cbk;
+	ent->context	= context;
+	ent->cmd	= cmd;
+	ent->page_queue = page_queue;
+
+	return ent;
+}
+
+static u8 alloc_token(struct mlx5_cmd *cmd)
+{
+	u8 token;
+
+	spin_lock(&cmd->token_spl);
+	token = cmd->token++ % 255 + 1;
+	spin_unlock(&cmd->token_spl);
+
+	return token;
+}
+
+static int alloc_ent(struct mlx5_cmd *cmd)
+{
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&cmd->alloc_spl, flags);
+	ret = find_first_bit(&cmd->bitmask, cmd->max_reg_cmds);
+	if (ret < cmd->max_reg_cmds)
+		clear_bit(ret, &cmd->bitmask);
+	spin_unlock_irqrestore(&cmd->alloc_spl, flags);
+
+	return ret < cmd->max_reg_cmds ? ret : -ENOMEM;
+}
+
+static void free_ent(struct mlx5_cmd *cmd, int idx)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&cmd->alloc_spl, flags);
+	set_bit(idx, &cmd->bitmask);
+	spin_unlock_irqrestore(&cmd->alloc_spl, flags);
+}
+
+static struct mlx5_cmd_layout *get_inst(struct mlx5_cmd *cmd, int idx)
+{
+	return cmd->cmd_buf + (idx << cmd->log_stride);
+}
+
+static u8 xor8_buf(void *buf, int len)
+{
+	u8 sum = 0;
+	int i;
+	u8 *ptr = buf;
+
+	for (i = 0; i < len; ++i)
+		sum ^= ptr[i];
+
+	return sum;
+}
+
+static int verify_block_sig(struct mlx5_cmd_prot_block *block)
+{
+	if (xor8_buf(block->rsvd0, sizeof(*block) - sizeof(block->data) - 1) != 0xff)
+		return -EINVAL;
+
+	if (xor8_buf(block, sizeof(*block)) != 0xff)
+		return -EINVAL;
+
+	return 0;
+}
+
+static void calc_block_sig(struct mlx5_cmd_prot_block *block, u8 token)
+{
+	block->token = token;
+	block->ctrl_sig = ~xor8_buf(block->rsvd0, sizeof(*block) - sizeof(block->data) - 2);
+	block->sig = ~xor8_buf(block, sizeof(*block) - 1);
+}
+
+static void calc_chain_sig(struct mlx5_cmd_msg *msg, u8 token)
+{
+	struct mlx5_cmd_mailbox *next = msg->next;
+
+	while (next) {
+		calc_block_sig(next->buf, token);
+		next = next->next;
+	}
+}
+
+static void set_signature(struct mlx5_cmd_work_ent *ent)
+{
+	ent->lay->sig = ~xor8_buf(ent->lay, sizeof(*ent->lay));
+	calc_chain_sig(ent->in, ent->token);
+	calc_chain_sig(ent->out, ent->token);
+}
+
+static void poll_timeout(struct mlx5_cmd_work_ent *ent)
+{
+	u8 own;
+	unsigned long poll_end = jiffies + msecs_to_jiffies(MLX5_CMD_TIMEOUT_MSEC + 1000);
+
+	do {
+		own = ent->lay->status_own;
+		if (!(own & CMD_OWNER_HW)) {
+			ent->ret = 0;
+			return;
+		}
+		usleep_range(5000, 10000);
+	} while (time_before(jiffies, poll_end));
+
+	ent->ret = -ETIMEDOUT;
+}
+
+static void free_cmd(struct mlx5_cmd_work_ent *ent)
+{
+	kfree(ent);
+}
+
+
+static int verify_signature(struct mlx5_cmd_work_ent *ent)
+{
+	int err;
+	u8 sig;
+	struct mlx5_cmd_mailbox *next = ent->out->next;
+
+	sig = xor8_buf(ent->lay, sizeof(*ent->lay));
+	if (sig != 0xff)
+		return -EINVAL;
+
+	while (next) {
+		err = verify_block_sig(next->buf);
+		if (err)
+			return err;
+
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static void dump_buf(void *buf, int size, int data_only, int offset)
+{
+	int i;
+	__be32 *p = buf;
+
+	if (!(debug_mask & mlx5_core_debug_mask) && !data_only)
+		return;
+
+	for (i = 0; i < size; i += 16) {
+		pr_debug("%03x: %08x %08x %08x %08x\n", offset, be32_to_cpu(p[0]),
+			 be32_to_cpu(p[1]), be32_to_cpu(p[2]),
+			 be32_to_cpu(p[3]));
+		p += 4;
+		offset += 16;
+	}
+	if (!data_only)
+		pr_debug("\n");
+}
+
+const char *mlx5_command_str(int command)
+{
+	switch (command) {
+	case MLX5_CMD_OP_QUERY_HCA_CAP:
+		return "QUERY_HCA_CAP";
+
+	case MLX5_CMD_OP_SET_HCA_CAP:
+		return "SET_HCA_CAP";
+
+	case MLX5_CMD_OP_QUERY_ADAPTER:
+		return "QUERY_ADAPTER";
+
+	case MLX5_CMD_OP_INIT_HCA:
+		return "INIT_HCA";
+
+	case MLX5_CMD_OP_TEARDOWN_HCA:
+		return "TEARDOWN_HCA";
+
+	case MLX5_CMD_OP_QUERY_PAGES:
+		return "QUERY_PAGES";
+
+	case MLX5_CMD_OP_MANAGE_PAGES:
+		return "MANAGE_PAGES";
+
+	case MLX5_CMD_OP_CREATE_MKEY:
+		return "CREATE_MKEY";
+
+	case MLX5_CMD_OP_QUERY_MKEY:
+		return "QUERY_MKEY";
+
+	case MLX5_CMD_OP_DESTROY_MKEY:
+		return "DESTROY_MKEY";
+
+	case MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS:
+		return "QUERY_SPECIAL_CONTEXTS";
+
+	case MLX5_CMD_OP_CREATE_EQ:
+		return "CREATE_EQ";
+
+	case MLX5_CMD_OP_DESTROY_EQ:
+		return "DESTROY_EQ";
+
+	case MLX5_CMD_OP_QUERY_EQ:
+		return "QUERY_EQ";
+
+	case MLX5_CMD_OP_CREATE_CQ:
+		return "CREATE_CQ";
+
+	case MLX5_CMD_OP_DESTROY_CQ:
+		return "DESTROY_CQ";
+
+	case MLX5_CMD_OP_QUERY_CQ:
+		return "QUERY_CQ";
+
+	case MLX5_CMD_OP_MODIFY_CQ:
+		return "MODIFY_CQ";
+
+	case MLX5_CMD_OP_CREATE_QP:
+		return "CREATE_QP";
+
+	case MLX5_CMD_OP_DESTROY_QP:
+		return "DESTROY_QP";
+
+	case MLX5_CMD_OP_RST2INIT_QP:
+		return "RST2INIT_QP";
+
+	case MLX5_CMD_OP_INIT2RTR_QP:
+		return "INIT2RTR_QP";
+
+	case MLX5_CMD_OP_RTR2RTS_QP:
+		return "RTR2RTS_QP";
+
+	case MLX5_CMD_OP_RTS2RTS_QP:
+		return "RTS2RTS_QP";
+
+	case MLX5_CMD_OP_SQERR2RTS_QP:
+		return "SQERR2RTS_QP";
+
+	case MLX5_CMD_OP_2ERR_QP:
+		return "2ERR_QP";
+
+	case MLX5_CMD_OP_RTS2SQD_QP:
+		return "RTS2SQD_QP";
+
+	case MLX5_CMD_OP_SQD2RTS_QP:
+		return "SQD2RTS_QP";
+
+	case MLX5_CMD_OP_2RST_QP:
+		return "2RST_QP";
+
+	case MLX5_CMD_OP_QUERY_QP:
+		return "QUERY_QP";
+
+	case MLX5_CMD_OP_CONF_SQP:
+		return "CONF_SQP";
+
+	case MLX5_CMD_OP_MAD_IFC:
+		return "MAD_IFC";
+
+	case MLX5_CMD_OP_INIT2INIT_QP:
+		return "INIT2INIT_QP";
+
+	case MLX5_CMD_OP_SUSPEND_QP:
+		return "SUSPEND_QP";
+
+	case MLX5_CMD_OP_UNSUSPEND_QP:
+		return "UNSUSPEND_QP";
+
+	case MLX5_CMD_OP_SQD2SQD_QP:
+		return "SQD2SQD_QP";
+
+	case MLX5_CMD_OP_ALLOC_QP_COUNTER_SET:
+		return "ALLOC_QP_COUNTER_SET";
+
+	case MLX5_CMD_OP_DEALLOC_QP_COUNTER_SET:
+		return "DEALLOC_QP_COUNTER_SET";
+
+	case MLX5_CMD_OP_QUERY_QP_COUNTER_SET:
+		return "QUERY_QP_COUNTER_SET";
+
+	case MLX5_CMD_OP_CREATE_PSV:
+		return "CREATE_PSV";
+
+	case MLX5_CMD_OP_DESTROY_PSV:
+		return "DESTROY_PSV";
+
+	case MLX5_CMD_OP_QUERY_PSV:
+		return "QUERY_PSV";
+
+	case MLX5_CMD_OP_QUERY_SIG_RULE_TABLE:
+		return "QUERY_SIG_RULE_TABLE";
+
+	case MLX5_CMD_OP_QUERY_BLOCK_SIZE_TABLE:
+		return "QUERY_BLOCK_SIZE_TABLE";
+
+	case MLX5_CMD_OP_CREATE_SRQ:
+		return "CREATE_SRQ";
+
+	case MLX5_CMD_OP_DESTROY_SRQ:
+		return "DESTROY_SRQ";
+
+	case MLX5_CMD_OP_QUERY_SRQ:
+		return "QUERY_SRQ";
+
+	case MLX5_CMD_OP_ARM_RQ:
+		return "ARM_RQ";
+
+	case MLX5_CMD_OP_RESIZE_SRQ:
+		return "RESIZE_SRQ";
+
+	case MLX5_CMD_OP_ALLOC_PD:
+		return "ALLOC_PD";
+
+	case MLX5_CMD_OP_DEALLOC_PD:
+		return "DEALLOC_PD";
+
+	case MLX5_CMD_OP_ALLOC_UAR:
+		return "ALLOC_UAR";
+
+	case MLX5_CMD_OP_DEALLOC_UAR:
+		return "DEALLOC_UAR";
+
+	case MLX5_CMD_OP_ATTACH_TO_MCG:
+		return "ATTACH_TO_MCG";
+
+	case MLX5_CMD_OP_DETACH_FROM_MCG:
+		return "DETACH_FROM_MCG";
+
+	case MLX5_CMD_OP_ALLOC_XRCD:
+		return "ALLOC_XRCD";
+
+	case MLX5_CMD_OP_DEALLOC_XRCD:
+		return "DEALLOC_XRCD";
+
+	case MLX5_CMD_OP_ACCESS_REG:
+		return "MLX5_CMD_OP_ACCESS_REG";
+
+	default: return "unknown command opcode";
+	}
+}
+
+static void dump_command(struct mlx5_core_dev *dev,
+			 struct mlx5_cmd_work_ent *ent, int input)
+{
+	struct mlx5_cmd_msg *msg = input ? ent->in : ent->out;
+	u16 op = be16_to_cpu(((struct mlx5_inbox_hdr *)(ent->lay->in))->opcode);
+	struct mlx5_cmd_mailbox *next = msg->next;
+	int data_only;
+	int mask = 1 << MLX5_MOD_CMD_DATA;
+	int offset = 0;
+	int dump_len;
+
+	data_only = !!(mlx5_core_debug_mask & mask);
+
+	if (!(debug_mask & mlx5_core_debug_mask) && !data_only)
+		return;
+
+	if (data_only)
+		mlx5_core_dbg_mask(dev, mask, "dump command data %s(0x%x) %s\n",
+				   mlx5_command_str(op), op,
+				   input ? "INPUT" : "OUTPUT");
+	else
+		mlx5_core_dbg(dev, "dump command %s(0x%x) %s\n",
+			      mlx5_command_str(op), op,
+			      input ? "INPUT" : "OUTPUT");
+
+	if (data_only) {
+		if (input) {
+			dump_buf(ent->lay->in, sizeof(ent->lay->in), 1, offset);
+			offset += sizeof(ent->lay->in);
+		} else {
+			dump_buf(ent->lay->out, sizeof(ent->lay->out), 1, offset);
+			offset += sizeof(ent->lay->out);
+		}
+	} else {
+		dump_buf(ent->lay, sizeof(*ent->lay), 0, offset);
+		offset += sizeof(*ent->lay);
+	}
+
+	while (next && offset < msg->len) {
+		if (data_only) {
+			dump_len = min_t(int, MLX5_CMD_DATA_BLOCK_SIZE, msg->len - offset);
+			dump_buf(next->buf, dump_len, 1, offset);
+			offset += MLX5_CMD_DATA_BLOCK_SIZE;
+		} else {
+			mlx5_core_dbg(dev, "command block:\n");
+			dump_buf(next->buf, sizeof(struct mlx5_cmd_prot_block), 0, offset);
+			offset += sizeof(struct mlx5_cmd_prot_block);
+		}
+		next = next->next;
+	}
+
+	if (data_only)
+		pr_debug("\n");
+}
+
+static void cmd_work_handler(struct work_struct *work)
+{
+	struct mlx5_cmd_work_ent *ent = container_of(work, struct mlx5_cmd_work_ent, work);
+	struct mlx5_cmd *cmd = ent->cmd;
+	struct mlx5_core_dev *dev = container_of(cmd, struct mlx5_core_dev, cmd);
+	struct mlx5_cmd_layout *lay;
+	struct semaphore *sem;
+
+	sem = ent->page_queue ? &cmd->pages_sem : &cmd->sem;
+	down(sem);
+	if (!ent->page_queue) {
+		ent->idx = alloc_ent(cmd);
+		if (ent->idx < 0) {
+			mlx5_core_err(dev, "failed to allocate command entry\n");
+			up(sem);
+			return;
+		}
+	} else {
+		ent->idx = cmd->max_reg_cmds;
+	}
+
+	ent->token = alloc_token(cmd);
+	cmd->ent_arr[ent->idx] = ent;
+	lay = get_inst(cmd, ent->idx);
+	ent->lay = lay;
+	memset(lay, 0, sizeof(*lay));
+	memcpy(lay->in, ent->in->first.data, sizeof(lay->in));
+	if (ent->in->next)
+		lay->in_ptr = cpu_to_be64(ent->in->next->dma);
+	lay->inlen = cpu_to_be32(ent->in->len);
+	if (ent->out->next)
+		lay->out_ptr = cpu_to_be64(ent->out->next->dma);
+	lay->outlen = cpu_to_be32(ent->out->len);
+	lay->type = MLX5_PCI_CMD_XPORT;
+	lay->token = ent->token;
+	lay->status_own = CMD_OWNER_HW;
+	if (!cmd->checksum_disabled)
+		set_signature(ent);
+	dump_command(dev, ent, 1);
+	ktime_get_ts(&ent->ts1);
+	/*
+	 * ring doorbell after the descriptor is valid
+	 */
+	wmb();
+	iowrite32be(1 << ent->idx, &dev->iseg->cmd_dbell);
+	mlx5_core_dbg(dev, "write 0x%x to command doorbell\n", 1 << ent->idx);
+	mmiowb();
+	if (cmd->mode == CMD_MODE_POLLING) {
+		poll_timeout(ent);
+		/*
+		 * make sure we read the descriptor after ownership is SW
+		 */
+		rmb();
+		mlx5_cmd_comp_handler(dev, 1UL << ent->idx);
+	}
+}
+
+static const char *status_to_str(u8 status)
+{
+	switch (status) {
+	case 0:
+		return "no errors";
+	case 1:
+		return "signature error";
+	case 2:
+		return "token error";
+	case 3:
+		return "bad block number";
+	case 4:
+		return "output pointer not aligned to block size";
+	case 5:
+		return "input pointer not aligned to block size";
+	case 6:
+		return "firmware internal error";
+	case 7:
+		return "command input length error";
+	case 8:
+		return "command ouput length error";
+	case 9:
+		return "reserved fields not cleared";
+	case 0x10:
+		return "bad command descriptor type";
+	default:
+		return "unknown status code\n";
+	}
+}
+
+static u16 msg_to_opcode(struct mlx5_cmd_msg *in)
+{
+	struct mlx5_inbox_hdr *hdr = (struct mlx5_inbox_hdr *)(in->first.data);
+
+	return be16_to_cpu(hdr->opcode);
+}
+
+static int wait_func(struct mlx5_core_dev *dev, struct mlx5_cmd_work_ent *ent)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int err;
+	unsigned long timeout = msecs_to_jiffies(MLX5_CMD_TIMEOUT_MSEC);
+
+	if (cmd->mode == CMD_MODE_POLLING) {
+		wait_for_completion(&ent->done);
+		err = ent->ret;
+	} else {
+		if (!wait_for_completion_timeout(&ent->done, timeout))
+			err = -ETIMEDOUT;
+		else
+			err = 0;
+	}
+	if (err == -ETIMEDOUT) {
+		mlx5_core_warn(dev, "%s(0x%x) timeout. Will cause a leak of a command resource\n",
+			       mlx5_command_str(msg_to_opcode(ent->in)),
+			       msg_to_opcode(ent->in));
+	}
+	mlx5_core_dbg(dev, "err %d, status %s(%d)\n", err,
+		      status_to_str(ent->status), ent->status);
+
+	return err;
+}
+
+/*  Notes:
+ *    1. Callback functions may not sleep
+ *    2. page queue commands do not support asynchrous completion
+ */
+static int mlx5_cmd_invoke(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *in,
+			   struct mlx5_cmd_msg *out, mlx5_cmd_cbk_t callback,
+			   void *context, int page_queue, u8 *status)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_work_ent *ent;
+	int err = 0;
+	ktime_t t1, t2, delta;
+	s64 ds;
+	u16 op;
+	struct mlx5_cmd_stats *stats;
+
+	if (callback && page_queue)
+		return -EINVAL;
+
+	ent = alloc_cmd(cmd, in, out, callback, context, page_queue);
+	if (IS_ERR(ent))
+		return PTR_ERR(ent);
+
+	if (!callback)
+		init_completion(&ent->done);
+
+	INIT_WORK(&ent->work, cmd_work_handler);
+	if (page_queue) {
+		cmd_work_handler(&ent->work);
+	} else if (!queue_work(cmd->wq, &ent->work)) {
+		mlx5_core_warn(dev, "failed to queue work\n");
+		err = -ENOMEM;
+		goto out_free;
+	}
+
+	if (!callback) {
+		err = wait_func(dev, ent);
+		if (err == -ETIMEDOUT)
+			goto out;
+
+		t1 = timespec_to_ktime(ent->ts1);
+		t2 = timespec_to_ktime(ent->ts2);
+		delta = ktime_sub(t2, t1);
+		ds = ktime_to_ns(delta);
+		op = be16_to_cpu(((struct mlx5_inbox_hdr *)in->first.data)->opcode);
+		if (op < ARRAY_SIZE(cmd->stats)) {
+			stats = &cmd->stats[op];
+			spin_lock(&stats->spl);
+			stats->sum += ds;
+			++stats->n;
+			spin_unlock(&stats->spl);
+		}
+		mlx5_core_dbg_mask(dev, 1 << MLX5_CMD_DATA_TIME,
+				   "fw exec time for %s is %lld nsec\n",
+				   mlx5_command_str(op), ds);
+		*status = ent->status;
+		free_cmd(ent);
+	}
+
+	return err;
+
+out_free:
+	free_cmd(ent);
+out:
+	return err;
+}
+
+static ssize_t dbg_write(struct file *filp, const char __user *buf,
+			 size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	int err;
+	char lbuf[3];
+
+	if (!dbg->in_msg || !dbg->out_msg)
+		return -ENOMEM;
+
+	if (copy_from_user(lbuf, buf, sizeof(lbuf)))
+		return -EPERM;
+
+	lbuf[sizeof(lbuf) - 1] = 0;
+
+	if (strcmp(lbuf, "go"))
+		return -EINVAL;
+
+	err = mlx5_cmd_exec(dev, dbg->in_msg, dbg->inlen, dbg->out_msg, dbg->outlen);
+
+	return err ? err : count;
+}
+
+
+static const struct file_operations fops = {
+	.owner	= THIS_MODULE,
+	.open	= dbg_open,
+	.write	= dbg_write,
+};
+
+static int mlx5_copy_to_msg(struct mlx5_cmd_msg *to, void *from, int size)
+{
+	int copy;
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_mailbox *next;
+
+	if (!to || !from)
+		return -ENOMEM;
+
+	copy = min_t(int, size, sizeof(to->first.data));
+	memcpy(to->first.data, from, copy);
+	size -= copy;
+	from += copy;
+
+	next = to->next;
+	while (size) {
+		if (!next) {
+			/* this is a BUG */
+			return -ENOMEM;
+		}
+
+		copy = min_t(int, size, MLX5_CMD_DATA_BLOCK_SIZE);
+		block = next->buf;
+		memcpy(block->data, from, copy);
+		from += copy;
+		size -= copy;
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static int mlx5_copy_from_msg(void *to, struct mlx5_cmd_msg *from, int size)
+{
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_mailbox *next;
+	int copy;
+
+	if (!to || !from)
+		return -ENOMEM;
+
+	copy = min_t(int, size, sizeof(from->first.data));
+	memcpy(to, from->first.data, copy);
+	size -= copy;
+	to += copy;
+
+	next = from->next;
+	while (size) {
+		if (!next) {
+			/* this is a BUG */
+			return -ENOMEM;
+		}
+
+		copy = min_t(int, size, MLX5_CMD_DATA_BLOCK_SIZE);
+		block = next->buf;
+		if (xor8_buf(block, sizeof(*block)) != 0xff)
+			return -EINVAL;
+
+		memcpy(to, block->data, copy);
+		to += copy;
+		size -= copy;
+		next = next->next;
+	}
+
+	return 0;
+}
+
+static struct mlx5_cmd_mailbox *alloc_cmd_box(struct mlx5_core_dev *dev,
+					      gfp_t flags)
+{
+	struct mlx5_cmd_mailbox *mailbox;
+
+	mailbox = kmalloc(sizeof(*mailbox), flags);
+	if (!mailbox) {
+		mlx5_core_dbg(dev, "failed allocation\n");
+		return ERR_PTR(-ENOMEM);
+	}
+
+	mailbox->buf = pci_pool_alloc(dev->cmd.pool, flags,
+				      &mailbox->dma);
+	if (!mailbox->buf) {
+		mlx5_core_dbg(dev, "failed allocation\n");
+		kfree(mailbox);
+		return ERR_PTR(-ENOMEM);
+	}
+	memset(mailbox->buf, 0, sizeof(struct mlx5_cmd_prot_block));
+	mailbox->next = NULL;
+
+	return mailbox;
+}
+
+static void free_cmd_box(struct mlx5_core_dev *dev,
+			 struct mlx5_cmd_mailbox *mailbox)
+{
+	pci_pool_free(dev->cmd.pool, mailbox->buf, mailbox->dma);
+	kfree(mailbox);
+}
+
+static struct mlx5_cmd_msg *mlx5_alloc_cmd_msg(struct mlx5_core_dev *dev,
+					       gfp_t flags, int size)
+{
+	struct mlx5_cmd_mailbox *tmp, *head = NULL;
+	struct mlx5_cmd_prot_block *block;
+	struct mlx5_cmd_msg *msg;
+	int n;
+	int i;
+	int blen;
+	int err;
+
+	msg = kzalloc(sizeof(*msg), GFP_KERNEL);
+	if (!msg)
+		return ERR_PTR(-ENOMEM);
+
+	blen = size - min_t(int, sizeof(msg->first.data), size);
+	n = (blen + MLX5_CMD_DATA_BLOCK_SIZE - 1) / MLX5_CMD_DATA_BLOCK_SIZE;
+
+	for (i = 0; i < n; ++i) {
+		tmp = alloc_cmd_box(dev, flags);
+		if (IS_ERR(tmp)) {
+			mlx5_core_warn(dev, "failed allocating block\n");
+			err = PTR_ERR(tmp);
+			goto err_alloc;
+		}
+
+		block = tmp->buf;
+		tmp->next = head;
+		block->next = cpu_to_be64(tmp->next ? tmp->next->dma : 0);
+		block->block_num = cpu_to_be32(n - i - 1);
+		head = tmp;
+	}
+	msg->next = head;
+	msg->len = size;
+	return msg;
+
+err_alloc:
+	while (head) {
+		tmp = head->next;
+		free_cmd_box(dev, head);
+		head = tmp;
+	}
+	kfree(msg);
+
+	return ERR_PTR(err);
+}
+
+static void mlx5_free_cmd_msg(struct mlx5_core_dev *dev,
+				  struct mlx5_cmd_msg *msg)
+{
+	struct mlx5_cmd_mailbox *head = msg->next;
+	struct mlx5_cmd_mailbox *next;
+
+	while (head) {
+		next = head->next;
+		free_cmd_box(dev, head);
+		head = next;
+	}
+	kfree(msg);
+}
+
+static ssize_t data_write(struct file *filp, const char __user *buf,
+			  size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	void *ptr;
+	int err;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+
+	if (*pos != 0)
+		return -EINVAL;
+
+	kfree(dbg->in_msg);
+	dbg->in_msg = NULL;
+	dbg->inlen = 0;
+
+	ptr = kzalloc(count, GFP_KERNEL);
+	if (!ptr)
+		return -ENOMEM;
+
+	if (copy_from_user(ptr, buf, count)) {
+		err = -EPERM;
+		goto out;
+	}
+	dbg->in_msg = ptr;
+	dbg->inlen = count;
+
+	*pos = count;
+
+	return count;
+
+out:
+	kfree(ptr);
+	return err;
+}
+
+static ssize_t data_read(struct file *filp, char __user *buf, size_t count,
+			 loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	int copy;
+
+	if (*pos)
+		return 0;
+
+	if (!dbg->out_msg)
+		return -ENOMEM;
+
+	copy = min_t(int, count, dbg->outlen);
+	if (copy_to_user(buf, dbg->out_msg, copy))
+		return -EPERM;
+
+	*pos += copy;
+
+	return copy;
+}
+
+static const struct file_operations dfops = {
+	.owner	= THIS_MODULE,
+	.open	= dbg_open,
+	.write	= data_write,
+	.read	= data_read,
+};
+
+static ssize_t outlen_read(struct file *filp, char __user *buf, size_t count,
+			   loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	int err;
+	char outlen[8];
+
+	if (*pos)
+		return 0;
+
+	err = snprintf(outlen, sizeof(outlen), "%d", dbg->outlen);
+	if (err < 0)
+		return err;
+
+	if (copy_to_user(buf, &outlen, err))
+		return -EPERM;
+
+	*pos += err;
+
+	return err;
+}
+
+static ssize_t outlen_write(struct file *filp, const char __user *buf,
+			    size_t count, loff_t *pos)
+{
+	struct mlx5_core_dev *dev = filp->private_data;
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	void *ptr;
+	int err;
+	char outlen_str[8];
+	int outlen;
+
+	if (*pos != 0 || count > 6)
+		return -EINVAL;
+
+	kfree(dbg->out_msg);
+	dbg->out_msg = NULL;
+	dbg->outlen = 0;
+
+	if (copy_from_user(outlen_str, buf, count))
+		return -EPERM;
+
+	outlen_str[7] = 0;
+
+	err = sscanf(outlen_str, "%d", &outlen);
+	if (err < 0)
+		return err;
+
+	ptr = kzalloc(outlen, GFP_KERNEL);
+	if (!ptr)
+		return -ENOMEM;
+
+	dbg->out_msg = ptr;
+	dbg->outlen = outlen;
+
+	*pos = count;
+
+	return count;
+}
+
+static const struct file_operations olfops = {
+	.owner	= THIS_MODULE,
+	.open	= dbg_open,
+	.write	= outlen_write,
+	.read	= outlen_read,
+};
+
+static void set_wqname(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+
+	strcpy(cmd->wq_name, "mlx5_cmd_");
+	strcat(cmd->wq_name, dev_name(&dev->pdev->dev));
+}
+
+static void clean_debug_files(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+
+	if (!mlx5_debugfs_root)
+		return;
+
+	mlx5_cmdif_debugfs_cleanup(dev);
+	debugfs_remove_recursive(dbg->dbg_root);
+}
+
+static int create_debugfs_files(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_debug *dbg = &dev->cmd.dbg;
+	int err = -ENOMEM;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	dbg->dbg_root = debugfs_create_dir("cmd", dev->priv.dbg_root);
+	if (!dbg->dbg_root)
+		return err;
+
+	dbg->dbg_in = debugfs_create_file("in", 0400, dbg->dbg_root,
+					  dev, &dfops);
+	if (!dbg->dbg_in)
+		goto err_dbg;
+
+	dbg->dbg_out = debugfs_create_file("out", 0200, dbg->dbg_root,
+					   dev, &dfops);
+	if (!dbg->dbg_out)
+		goto err_dbg;
+
+	dbg->dbg_outlen = debugfs_create_file("out_len", 0600, dbg->dbg_root,
+					      dev, &olfops);
+	if (!dbg->dbg_outlen)
+		goto err_dbg;
+
+	dbg->dbg_status = debugfs_create_u8("status", 0600, dbg->dbg_root,
+					    &dbg->status);
+	if (!dbg->dbg_status)
+		goto err_dbg;
+
+	dbg->dbg_run = debugfs_create_file("run", 0200, dbg->dbg_root, dev, &fops);
+	if (!dbg->dbg_run)
+		goto err_dbg;
+
+	mlx5_cmdif_debugfs_init(dev);
+
+	return 0;
+
+err_dbg:
+	clean_debug_files(dev);
+	return err;
+}
+
+void mlx5_cmd_use_events(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	for (i = 0; i < cmd->max_reg_cmds; ++i)
+		down(&cmd->sem);
+
+	down(&cmd->pages_sem);
+
+	flush_workqueue(cmd->wq);
+
+	cmd->mode = CMD_MODE_EVENTS;
+
+	up(&cmd->pages_sem);
+	for (i = 0; i < cmd->max_reg_cmds; ++i)
+		up(&cmd->sem);
+}
+
+void mlx5_cmd_use_polling(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	int i;
+
+	for (i = 0; i < cmd->max_reg_cmds; ++i)
+		down(&cmd->sem);
+
+	down(&cmd->pages_sem);
+
+	flush_workqueue(cmd->wq);
+	cmd->mode = CMD_MODE_POLLING;
+
+	up(&cmd->pages_sem);
+	for (i = 0; i < cmd->max_reg_cmds; ++i)
+		up(&cmd->sem);
+}
+
+void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_work_ent *ent;
+	int i;
+	mlx5_cmd_cbk_t callback;
+	int err;
+	void *context;
+
+	for (i = 0; i < (1 << cmd->log_sz); ++i) {
+		if (test_bit(i, &vector)) {
+			ent = cmd->ent_arr[i];
+			ktime_get_ts(&ent->ts2);
+			memcpy(ent->out->first.data, ent->lay->out, sizeof(ent->lay->out));
+			dump_command(dev, ent, 0);
+			if (!ent->ret) {
+				if (!cmd->checksum_disabled)
+					ent->ret = verify_signature(ent);
+				else
+					ent->ret = 0;
+				ent->status = ent->lay->status_own >> 1;
+				mlx5_core_dbg(dev, "command completed. ret 0x%x, status %s(0x%x)\n", ent->ret,
+					      status_to_str(ent->status), ent->status);
+			}
+			free_ent(cmd, ent->idx);
+			if (ent->callback) {
+				callback = ent->callback;
+				context = ent->context;
+				err = ent->ret;
+				free_cmd(ent);
+				callback(err, context);
+			} else {
+				complete(&ent->done);
+			}
+			if (ent->page_queue)
+				up(&cmd->pages_sem);
+			else
+				up(&cmd->sem);
+		}
+	}
+}
+EXPORT_SYMBOL(mlx5_cmd_comp_handler);
+
+static int status_to_err(u8 status)
+{
+	return status ? -1 : 0; /* TBD more meaningful codes */
+}
+
+static struct mlx5_cmd_msg *alloc_msg(struct mlx5_core_dev *dev, int in_size)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_msg *msg = ERR_PTR(-ENOMEM);
+	struct cache_ent *ent = NULL;
+
+	if (in_size > MED_LIST_SIZE && in_size <= LONG_LIST_SIZE)
+		ent = &cmd->cache.large;
+	else if (in_size > 16 && in_size <= MED_LIST_SIZE)
+		ent = &cmd->cache.med;
+
+	if (ent) {
+		spin_lock(&ent->lock);
+		if (!list_empty(&ent->head)) {
+			msg = list_entry(ent->head.next, typeof(*msg), list);
+			/*
+			 * for cached lists, we must explicitly state what is
+			 * the real size
+			 */
+			msg->len = in_size;
+			list_del(&msg->list);
+		}
+		spin_unlock(&ent->lock);
+	}
+
+	if (IS_ERR(msg))
+		msg = mlx5_alloc_cmd_msg(dev, GFP_KERNEL, in_size);
+
+	return msg;
+}
+
+static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg)
+{
+	if (msg->cache) {
+		spin_lock(&msg->cache->lock);
+		list_add_tail(&msg->list, &msg->cache->head);
+		spin_unlock(&msg->cache->lock);
+	} else {
+		mlx5_free_cmd_msg(dev, msg);
+	}
+}
+
+static int is_manage_pages(struct mlx5_inbox_hdr *in)
+{
+	return be16_to_cpu(in->opcode) == MLX5_CMD_OP_MANAGE_PAGES;
+}
+
+int mlx5_cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out,
+		  int out_size)
+{
+	struct mlx5_cmd_msg *inb;
+	struct mlx5_cmd_msg *outb;
+	int err;
+	u8 status = 0;
+	int pages_queue;
+
+	pages_queue = is_manage_pages(in);
+
+	inb = alloc_msg(dev, in_size);
+	if (IS_ERR(inb)) {
+		err = PTR_ERR(inb);
+		return err;
+	}
+
+	err = mlx5_copy_to_msg(inb, in, in_size);
+	if (err) {
+		mlx5_core_warn(dev, "err %d\n", err);
+		goto out_in;
+	}
+
+	outb = mlx5_alloc_cmd_msg(dev, GFP_KERNEL, out_size);
+	if (IS_ERR(outb)) {
+		err = PTR_ERR(outb);
+		goto out_in;
+	}
+
+	err = mlx5_cmd_invoke(dev, inb, outb, NULL, NULL, pages_queue, &status);
+	if (err)
+		goto out_out;
+
+	mlx5_core_dbg(dev, "err %d, status %d\n", err, status);
+	if (status) {
+		err = status_to_err(status);
+		goto out_out;
+	}
+
+	err = mlx5_copy_from_msg(out, outb, out_size);
+
+out_out:
+	mlx5_free_cmd_msg(dev, outb);
+
+out_in:
+	free_msg(dev, inb);
+	return err;
+}
+EXPORT_SYMBOL(mlx5_cmd_exec);
+
+static int create_msg_cache(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_msg *msg;
+	struct mlx5_cmd_msg *n;
+	int err;
+	int i;
+
+	spin_lock_init(&cmd->cache.large.lock);
+	INIT_LIST_HEAD(&cmd->cache.large.head);
+	spin_lock_init(&cmd->cache.med.lock);
+	INIT_LIST_HEAD(&cmd->cache.med.head);
+
+	for (i = 0; i < NUM_LONG_LISTS; ++i) {
+		msg = mlx5_alloc_cmd_msg(dev, GFP_KERNEL, LONG_LIST_SIZE);
+		if (IS_ERR(msg)) {
+			err = PTR_ERR(msg);
+			goto ex_err;
+		}
+		msg->cache = &cmd->cache.large;
+		list_add_tail(&msg->list, &cmd->cache.large.head);
+	}
+
+	for (i = 0; i < NUM_MED_LISTS; ++i) {
+		msg = mlx5_alloc_cmd_msg(dev, GFP_KERNEL, MED_LIST_SIZE);
+		if (IS_ERR(msg)) {
+			err = PTR_ERR(msg);
+			goto ex_err;
+		}
+		msg->cache = &cmd->cache.med;
+		list_add_tail(&msg->list, &cmd->cache.med.head);
+	}
+
+	return 0;
+
+ex_err:
+	list_for_each_entry_safe(msg, n, &cmd->cache.large.head, list) {
+		list_del(&msg->list);
+		mlx5_free_cmd_msg(dev, msg);
+	}
+
+	list_for_each_entry_safe(msg, n, &cmd->cache.med.head, list) {
+		list_del(&msg->list);
+		mlx5_free_cmd_msg(dev, msg);
+	}
+
+	return err;
+}
+
+static void destroy_msg_cache(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	struct mlx5_cmd_msg *msg;
+	struct mlx5_cmd_msg *n;
+
+	list_for_each_entry_safe(msg, n, &cmd->cache.large.head, list) {
+		list_del(&msg->list);
+		mlx5_free_cmd_msg(dev, msg);
+	}
+
+	list_for_each_entry_safe(msg, n, &cmd->cache.med.head, list) {
+		list_del(&msg->list);
+		mlx5_free_cmd_msg(dev, msg);
+	}
+}
+
+int mlx5_cmd_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+	u32 cmd_h, cmd_l;
+	int err;
+	int size = sizeof(struct mlx5_cmd_prot_block);
+	int align = roundup_pow_of_two(size);
+	u16 cmd_if_rev;
+	int i;
+
+	cmd_if_rev = cmdif_rev(dev);
+	if (cmd_if_rev != CMD_IF_REV) {
+		dev_err(&dev->pdev->dev,
+			"Driver cmdif rev(%d) differs from firmware's(%d)\n",
+			CMD_IF_REV, cmd_if_rev);
+		return -EINVAL;
+	}
+
+	cmd->pool = pci_pool_create("mlx5_cmd", dev->pdev, size, align, 0);
+	if (!cmd->pool)
+		return -ENOMEM;
+
+	cmd->cmd_buf = (void *)__get_free_pages(GFP_ATOMIC, 0);
+	if (!cmd->cmd_buf) {
+		err = -ENOMEM;
+		goto err_free_pool;
+	}
+	cmd->dma = dma_map_single(&dev->pdev->dev, cmd->cmd_buf, PAGE_SIZE,
+				  DMA_BIDIRECTIONAL);
+	if (dma_mapping_error(&dev->pdev->dev, cmd->dma)) {
+		err = -ENOMEM;
+		goto err_free;
+	}
+
+	cmd_l = ioread32be(&dev->iseg->cmdq_addr_l_sz) & 0xff;
+	cmd->log_sz = cmd_l >> 4 & 0xf;
+	cmd->log_stride = cmd_l & 0xf;
+	if (1 << cmd->log_sz > MLX5_MAX_COMMANDS) {
+		dev_err(&dev->pdev->dev, "firmware reports too many outstanding commands %d\n",
+			1 << cmd->log_sz);
+		err = -EINVAL;
+		goto err_map;
+	}
+
+	if (cmd->log_sz + cmd->log_stride > PAGE_SHIFT) {
+		dev_err(&dev->pdev->dev, "command queue size overflow\n");
+		err = -EINVAL;
+		goto err_map;
+	}
+
+	cmd->max_reg_cmds = (1 << cmd->log_sz) - 1;
+	cmd->bitmask = (1 << cmd->max_reg_cmds) - 1;
+
+	cmd->cmdif_rev = ioread32be(&dev->iseg->cmdif_rev_fw_sub) >> 16;
+	if (cmd->cmdif_rev > CMD_IF_REV) {
+		dev_err(&dev->pdev->dev, "driver does not support command interface version. driver %d, firmware %d\n",
+			CMD_IF_REV, cmd->cmdif_rev);
+		err = -ENOTSUPP;
+		goto err_map;
+	}
+
+	spin_lock_init(&cmd->alloc_spl);
+	spin_lock_init(&cmd->token_spl);
+	for (i = 0; i < ARRAY_SIZE(cmd->stats); ++i)
+		spin_lock_init(&cmd->stats[i].spl);
+
+	sema_init(&cmd->sem, cmd->max_reg_cmds);
+	sema_init(&cmd->pages_sem, 1);
+
+	cmd_h = (u32)((u64)(cmd->dma) >> 32);
+	cmd_l = (u32)(cmd->dma);
+	if (cmd_l & 0xfff) {
+		dev_err(&dev->pdev->dev, "invalid command queue address\n");
+		err = -ENOMEM;
+		goto err_map;
+	}
+
+	iowrite32be(cmd_h, &dev->iseg->cmdq_addr_h);
+	iowrite32be(cmd_l, &dev->iseg->cmdq_addr_l_sz);
+	/*
+	 * make sure firmware sees the complete address before we proceed
+	 */
+	wmb();
+
+	mlx5_core_dbg(dev, "descriptor at dma 0x%llx\n", (unsigned long long)(cmd->dma));
+
+	cmd->mode = CMD_MODE_POLLING;
+
+	err = create_msg_cache(dev);
+	if (err) {
+		dev_err(&dev->pdev->dev, "failed to create command cache\n");
+		goto err_map;
+	}
+
+	set_wqname(dev);
+	cmd->wq = create_singlethread_workqueue(cmd->wq_name);
+	if (!cmd->wq) {
+		dev_err(&dev->pdev->dev, "failed to create command workqueue\n");
+		err = -ENOMEM;
+		goto err_cache;
+	}
+
+	err = create_debugfs_files(dev);
+	if (err) {
+		err = -ENOMEM;
+		goto err_wq;
+	}
+
+	return 0;
+
+err_wq:
+	destroy_workqueue(cmd->wq);
+
+err_cache:
+	destroy_msg_cache(dev);
+
+err_map:
+	dma_unmap_single(&dev->pdev->dev, cmd->dma, PAGE_SIZE,
+			 DMA_BIDIRECTIONAL);
+err_free:
+	free_pages((unsigned long)cmd->cmd_buf, 0);
+
+err_free_pool:
+	pci_pool_destroy(cmd->pool);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_cmd_init);
+
+void mlx5_cmd_cleanup(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd *cmd = &dev->cmd;
+
+	clean_debug_files(dev);
+	destroy_workqueue(cmd->wq);
+	destroy_msg_cache(dev);
+	dma_unmap_single(&dev->pdev->dev, cmd->dma, PAGE_SIZE,
+			 DMA_BIDIRECTIONAL);
+	free_pages((unsigned long)cmd->cmd_buf, 0);
+	pci_pool_destroy(cmd->pool);
+}
+EXPORT_SYMBOL(mlx5_cmd_cleanup);
+
+static const char *cmd_status_str(u8 status)
+{
+	switch (status) {
+	case 0x0:	return "OK";
+	case 0x1:	return "internal error";
+	case 0x2:	return "bad operation";
+	case 0x3:	return "bad parameter";
+	case 0x4:	return "bad system state";
+	case 0x5:	return "bad resource";
+	case 0x6:	return "resource busy";
+	case 0x8:	return "limits exceeded";
+	case 0x9:	return "bad resource state";
+	case 0xa:	return "bad index";
+	case 0xf:	return "no resources";
+	case 0x50:	return "bad input length";
+	case 0x51:	return "bad output length";
+	case 0x10:	return "bad QP state";
+	case 0x30:	return "bad packet (discarded)";
+	case 0x40:	return "bad size too many outstanding CQEs";
+	default:	return "unknown status";
+	}
+}
+
+int mlx5_cmd_status_to_err(struct mlx5_outbox_hdr *hdr)
+{
+	if (!hdr->status)
+		return 0;
+
+	pr_warn("command failed, status %s(0x%x), syndrome 0x%x\n",
+		cmd_status_str(hdr->status), hdr->status,
+		be32_to_cpu(hdr->syndrome));
+
+	switch (hdr->status) {
+	case 0x0:	return 0;
+	case 0x1:	return -EIO;
+	case 0x2:	return -EINVAL;
+	case 0x3:	return -EINVAL;
+	case 0x4:	return -EIO;
+	case 0x5:	return -EINVAL;
+	case 0x6:	return -EBUSY;
+	case 0x8:	return -EINVAL;
+	case 0x9:	return -EINVAL;
+	case 0xa:	return -EINVAL;
+	case 0xf:	return -EAGAIN;
+	case 0x50:	return -EIO;
+	case 0x51:	return -EIO;
+	case 0x10:	return -EINVAL;
+	case 0x30:	return -EINVAL;
+	case 0x40:	return -EINVAL;
+	default:	return -EIO;
+	}
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cq.c b/drivers/net/ethernet/mellanox/mlx5/core/cq.c
new file mode 100644
index 0000000..7ab5f32
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cq.c
@@ -0,0 +1,226 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/hardirq.h>
+#include <linux/mlx5/driver.h>
+#include <linux/mlx5/cmd.h>
+#include <rdma/ib_verbs.h>
+#include <linux/mlx5/cq.h>
+#include "mlx5_core.h"
+
+MLX5_MOD_DBG_MASK(MLX5_MOD_CQ);
+
+void mlx5_cq_completion(struct mlx5_core_dev *dev, u32 cqn)
+{
+	struct mlx5_core_cq *cq;
+	struct mlx5_cq_table *table = &dev->priv.cq_table;
+
+	spin_lock(&table->lock);
+	cq = radix_tree_lookup(&table->tree, cqn);
+	if (likely(cq))
+		atomic_inc(&cq->refcount);
+	spin_unlock(&table->lock);
+
+	if (!cq) {
+		mlx5_core_warn(dev, "Completion event for bogus CQ 0x%x\n", cqn);
+		return;
+	}
+
+	++cq->arm_sn;
+
+	cq->comp(cq);
+
+	if (atomic_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+}
+
+void mlx5_cq_event(struct mlx5_core_dev *dev, u32 cqn, int event_type)
+{
+	struct mlx5_cq_table *table = &dev->priv.cq_table;
+	struct mlx5_core_cq *cq;
+
+	spin_lock(&table->lock);
+
+	cq = radix_tree_lookup(&table->tree, cqn);
+	if (cq)
+		atomic_inc(&cq->refcount);
+
+	spin_unlock(&table->lock);
+
+	if (!cq) {
+		mlx5_core_warn(dev, "Async event for bogus CQ 0x%x\n", cqn);
+		return;
+	}
+
+	cq->event(cq, event_type);
+
+	if (atomic_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+}
+
+
+int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			struct mlx5_create_cq_mbox_in *in, int inlen)
+{
+	int err;
+	struct mlx5_cq_table *table = &dev->priv.cq_table;
+	struct mlx5_create_cq_mbox_out out;
+	struct mlx5_destroy_cq_mbox_in din;
+	struct mlx5_destroy_cq_mbox_out dout;
+
+	in->hdr.opcode = cpu_to_be16(MLX5_CMD_OP_CREATE_CQ);
+	memset(&out, 0, sizeof(out));
+	err = mlx5_cmd_exec(dev, in, inlen, &out, sizeof(out));
+	if (err)
+		return err;
+
+	if (out.hdr.status)
+		return mlx5_cmd_status_to_err(&out.hdr);
+
+	cq->cqn = be32_to_cpu(out.cqn) & 0xffffff;
+	cq->cons_index = 0;
+	cq->arm_sn     = 0;
+	atomic_set(&cq->refcount, 1);
+	init_completion(&cq->free);
+
+	spin_lock_irq(&table->lock);
+	err = radix_tree_insert(&table->tree, cq->cqn, cq);
+	spin_unlock_irq(&table->lock);
+	if (err)
+		goto err_cmd;
+
+	cq->pid = current->pid;
+	err = mlx5_debug_cq_add(dev, cq);
+	if (err)
+		mlx5_core_dbg(dev, "failed adding CP 0x%x to debug file system\n",
+			      cq->cqn);
+
+	return 0;
+
+err_cmd:
+	memset(&din, 0, sizeof(din));
+	memset(&dout, 0, sizeof(dout));
+	din.hdr.opcode = cpu_to_be16(MLX5_CMD_OP_DESTROY_CQ);
+	mlx5_cmd_exec(dev, &din, sizeof(din), &dout, sizeof(dout));
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_create_cq);
+
+int mlx5_core_destroy_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	struct mlx5_destroy_cq_mbox_in in;
+	struct mlx5_destroy_cq_mbox_out out;
+	int err;
+	struct mlx5_cq_table *table = &dev->priv.cq_table;
+	struct mlx5_core_cq *tmp;
+
+	spin_lock_irq(&table->lock);
+	tmp = radix_tree_delete(&table->tree, cq->cqn);
+	spin_unlock_irq(&table->lock);
+	if (!tmp) {
+		mlx5_core_warn(dev, "cq 0x%x not found in tree\n", cq->cqn);
+		return -EINVAL;
+	}
+	if (tmp != cq) {
+		mlx5_core_warn(dev, "corruption on srqn 0x%x\n", cq->cqn);
+		return -EINVAL;
+	}
+
+	memset(&in, 0, sizeof(in));
+	memset(&out, 0, sizeof(out));
+	in.hdr.opcode = cpu_to_be16(MLX5_CMD_OP_DESTROY_CQ);
+	in.cqn = cpu_to_be32(cq->cqn);
+	err = mlx5_cmd_exec(dev, &in, sizeof(in), &out, sizeof(out));
+	if (err)
+		return err;
+
+	if (out.hdr.status)
+		return mlx5_cmd_status_to_err(&out.hdr);
+
+	synchronize_irq(cq->irqn);
+
+	mlx5_debug_cq_remove(dev, cq);
+	if (atomic_dec_and_test(&cq->refcount))
+		complete(&cq->free);
+	wait_for_completion(&cq->free);
+
+	return 0;
+}
+EXPORT_SYMBOL(mlx5_core_destroy_cq);
+
+int mlx5_core_query_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+		       struct mlx5_query_cq_mbox_out *out)
+{
+	struct mlx5_query_cq_mbox_in in;
+	int err;
+
+	memset(&in, 0, sizeof(in));
+	memset(out, 0, sizeof(*out));
+
+	in.hdr.opcode = cpu_to_be16(MLX5_CMD_OP_QUERY_CQ);
+	in.cqn = cpu_to_be32(cq->cqn);
+	err = mlx5_cmd_exec(dev, &in, sizeof(in), out, sizeof(*out));
+	if (err)
+		return err;
+
+	if (out->hdr.status)
+		return mlx5_cmd_status_to_err(&out->hdr);
+
+	return err;
+}
+EXPORT_SYMBOL(mlx5_core_query_cq);
+
+
+int mlx5_core_modify_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			int type, struct mlx5_cq_modify_params *params)
+{
+	return -ENOSYS;
+}
+
+int mlx5_init_cq_table(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cq_table *table = &dev->priv.cq_table;
+	int err;
+
+	spin_lock_init(&table->lock);
+	INIT_RADIX_TREE(&table->tree, GFP_ATOMIC);
+	err = mlx5_cq_debugfs_init(dev);
+
+	return err;
+}
+
+void mlx5_cleanup_cq_table(struct mlx5_core_dev *dev)
+{
+	mlx5_cq_debugfs_cleanup(dev);
+}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
new file mode 100644
index 0000000..f22113e
--- /dev/null
+++ b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c
@@ -0,0 +1,600 @@ 
+/*
+ * Copyright (c) 2013, Mellanox Technologies inc.  All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/module.h>
+#include <linux/debugfs.h>
+#include <linux/mlx5/qp.h>
+#include <linux/mlx5/cq.h>
+#include <linux/mlx5/driver.h>
+#include "mlx5_core.h"
+
+MLX5_MOD_DBG_MASK(MLX5_MOD_DEBUG);
+
+enum {
+	QP_PID,
+	QP_STATE,
+	QP_XPORT,
+	QP_MTU,
+	QP_N_RECV,
+	QP_RECV_SZ,
+	QP_N_SEND,
+	QP_LOG_PG_SZ,
+	QP_RQPN,
+};
+
+static char *qp_fields[] = {
+	[QP_PID]	= "pid",
+	[QP_STATE]	= "state",
+	[QP_XPORT]	= "transport",
+	[QP_MTU]	= "mtu",
+	[QP_N_RECV]	= "num_recv",
+	[QP_RECV_SZ]	= "rcv_wqe_sz",
+	[QP_N_SEND]	= "num_send",
+	[QP_LOG_PG_SZ]	= "log2_page_sz",
+	[QP_RQPN]	= "remote_qpn",
+};
+
+enum {
+	EQ_NUM_EQES,
+	EQ_INTR,
+	EQ_LOG_PG_SZ,
+};
+
+static char *eq_fields[] = {
+	[EQ_NUM_EQES]	= "num_eqes",
+	[EQ_INTR]	= "intr",
+	[EQ_LOG_PG_SZ]	= "log_page_size",
+};
+
+enum {
+	CQ_PID,
+	CQ_NUM_CQES,
+	CQ_LOG_PG_SZ,
+};
+
+static char *cq_fields[] = {
+	[CQ_PID]	= "pid",
+	[CQ_NUM_CQES]	= "num_cqes",
+	[CQ_LOG_PG_SZ]	= "log_page_size",
+};
+
+struct dentry *mlx5_debugfs_root;
+EXPORT_SYMBOL(mlx5_debugfs_root);
+
+void mlx5_register_debugfs(void)
+{
+	mlx5_debugfs_root = debugfs_create_dir("mlx5", NULL);
+	if (IS_ERR_OR_NULL(mlx5_debugfs_root))
+		mlx5_debugfs_root = NULL;
+}
+
+void mlx5_unregister_debugfs(void)
+{
+	debugfs_remove(mlx5_debugfs_root);
+}
+
+int mlx5_qp_debugfs_init(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	atomic_set(&dev->num_qps, 0);
+
+	dev->priv.qp_debugfs = debugfs_create_dir("QPs",  dev->priv.dbg_root);
+	if (!dev->priv.qp_debugfs)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx5_qp_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	debugfs_remove_recursive(dev->priv.qp_debugfs);
+}
+
+int mlx5_eq_debugfs_init(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	dev->priv.eq_debugfs = debugfs_create_dir("EQs",  dev->priv.dbg_root);
+	if (!dev->priv.eq_debugfs)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	debugfs_remove_recursive(dev->priv.eq_debugfs);
+}
+
+static int dbg_open(struct inode *inode, struct file *file)
+{
+	/*
+	 * inode.i_private is equal to the data argument passed to
+	 * debugfs_create_file
+	 */
+	file->private_data = inode->i_private;
+
+	return 0;
+}
+
+static ssize_t average_read(struct file *filp, char __user *buf, size_t count,
+			    loff_t *pos)
+{
+	struct mlx5_cmd_stats *stats;
+	u64 field = 0;
+	int ret;
+	int err;
+	char tbuf[22];
+
+	if (*pos)
+		return 0;
+
+	stats = filp->private_data;
+	spin_lock(&stats->spl);
+	if (stats->n)
+		field = stats->sum / stats->n;
+	spin_unlock(&stats->spl);
+	ret = snprintf(tbuf, sizeof(tbuf), "%llu\n", field);
+	if (ret > 0) {
+		err = copy_to_user(buf, tbuf, ret);
+		if (err)
+			return err;
+	}
+
+	*pos += ret;
+	return ret;
+}
+
+
+static ssize_t average_write(struct file *filp, const char __user *buf,
+			     size_t count, loff_t *pos)
+{
+	struct mlx5_cmd_stats *stats;
+
+	stats = filp->private_data;
+	spin_lock(&stats->spl);
+	stats->sum = 0;
+	stats->n = 0;
+	spin_unlock(&stats->spl);
+
+	*pos += count;
+
+	return count;
+}
+
+static const struct file_operations stats_fops = {
+	.owner	= THIS_MODULE,
+	.open	= dbg_open,
+	.read	= average_read,
+	.write	= average_write,
+};
+
+int mlx5_cmdif_debugfs_init(struct mlx5_core_dev *dev)
+{
+	struct mlx5_cmd_stats *stats;
+	int i;
+	const char *namep;
+	int err;
+	struct dentry **cmd;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	cmd = &dev->priv.cmdif_debugfs;
+	*cmd = debugfs_create_dir("commands", dev->priv.dbg_root);
+	if (!*cmd)
+		return -ENOMEM;
+
+	for (i = 0; i < ARRAY_SIZE(dev->cmd.stats); ++i) {
+		stats = &dev->cmd.stats[i];
+		namep = mlx5_command_str(i);
+		if (strcmp(namep, "unknown command opcode")) {
+			stats->root = debugfs_create_dir(namep, *cmd);
+			if (!stats->root) {
+				mlx5_core_warn(dev, "failed adding command %d\n",
+					       i);
+				err = -ENOMEM;
+				goto out;
+			}
+
+			stats->avg = debugfs_create_file("average", 0400,
+							 stats->root, stats,
+							 &stats_fops);
+			if (!stats->avg) {
+				mlx5_core_warn(dev, "failed creating debugfs file\n");
+				err = -ENOMEM;
+				goto out;
+			}
+
+			stats->count = debugfs_create_u64("n", 0400,
+							  stats->root,
+							  &stats->n);
+			if (!stats->count) {
+				mlx5_core_warn(dev, "failed creating debugfs file\n");
+				err = -ENOMEM;
+				goto out;
+			}
+		}
+	}
+
+	return 0;
+out:
+	debugfs_remove_recursive(dev->priv.cmdif_debugfs);
+	return err;
+}
+
+void mlx5_cmdif_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	debugfs_remove_recursive(dev->priv.cmdif_debugfs);
+}
+
+int mlx5_cq_debugfs_init(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	dev->priv.cq_debugfs = debugfs_create_dir("CQs",  dev->priv.dbg_root);
+	if (!dev->priv.cq_debugfs)
+		return -ENOMEM;
+
+	return 0;
+}
+
+void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	debugfs_remove_recursive(dev->priv.cq_debugfs);
+}
+
+static u64 qp_read_field(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp,
+			 int index)
+{
+	struct mlx5_query_qp_mbox_out *out;
+	struct mlx5_qp_context *ctx;
+	int err;
+	u64 param = 0;
+	int no_sq;
+
+	out = kzalloc(sizeof(*out), GFP_KERNEL);
+	if (!out)
+		return param;
+
+	err = mlx5_core_qp_query(dev, qp, out, sizeof(*out));
+	if (err) {
+		mlx5_core_warn(dev, "failed to query qp\n");
+		goto out;
+	}
+
+	ctx = &out->ctx;
+	switch (index) {
+	case QP_PID:
+		param = qp->pid;
+		break;
+	case QP_STATE:
+		param = be32_to_cpu(ctx->flags) >> 28;
+		break;
+	case QP_XPORT:
+		param = (be32_to_cpu(ctx->flags) >> 16) & 0xff;
+		break;
+	case QP_MTU:
+		param = ctx->mtu_msgmax >> 5;
+		break;
+	case QP_N_RECV:
+		param = 1 << ((ctx->rq_size_stride >> 3) & 0xf);
+		break;
+	case QP_RECV_SZ:
+		param = 1 << ((ctx->rq_size_stride & 7) + 4);
+		break;
+	case QP_N_SEND:
+		no_sq = be16_to_cpu(ctx->sq_crq_size) >> 15;
+		if (!no_sq)
+			param = 1 << (be16_to_cpu(ctx->sq_crq_size) >> 11);
+		else
+			param = 0;
+		break;
+	case QP_LOG_PG_SZ:
+		param = ((cpu_to_be32(ctx->log_pg_sz_remote_qpn) >> 24) & 0x1f);
+		param += 12;
+		break;
+	case QP_RQPN:
+		param = cpu_to_be32(ctx->log_pg_sz_remote_qpn) & 0xffffff;
+		break;
+	}
+
+out:
+	kfree(out);
+	return param;
+}
+
+static u64 eq_read_field(struct mlx5_core_dev *dev, struct mlx5_eq *eq,
+			 int index)
+{
+	struct mlx5_query_eq_mbox_out *out;
+	struct mlx5_eq_context *ctx;
+	int err;
+	u64 param = 0;
+
+	out = kzalloc(sizeof(*out), GFP_KERNEL);
+	if (!out)
+		return param;
+
+	ctx = &out->ctx;
+
+	err = mlx5_core_eq_query(dev, eq, out, sizeof(*out));
+	if (err) {
+		mlx5_core_warn(dev, "failed to query eq\n");
+		goto out;
+	}
+
+	switch (index) {
+	case EQ_NUM_EQES:
+		param = 1 << ((be32_to_cpu(ctx->log_sz_usr_page) >> 24) & 0x1f);
+		break;
+	case EQ_INTR:
+		param = ctx->intr;
+		break;
+	case EQ_LOG_PG_SZ:
+		param = (ctx->log_page_size & 0x1f) + 12;
+		break;
+	}
+
+out:
+	kfree(out);
+	return param;
+}
+
+static u64 cq_read_field(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq,
+			 int index)
+{
+	struct mlx5_query_cq_mbox_out *out;
+	struct mlx5_cq_context *ctx;
+	int err;
+	u64 param = 0;
+
+	out = kzalloc(sizeof(*out), GFP_KERNEL);
+	if (!out)
+		return param;
+
+	ctx = &out->ctx;
+
+	err = mlx5_core_query_cq(dev, cq, out);
+	if (err) {
+		mlx5_core_warn(dev, "failed to query cq\n");
+		goto out;
+	}
+
+	switch (index) {
+	case CQ_PID:
+		param = cq->pid;
+		break;
+	case CQ_NUM_CQES:
+		param = 1 << ((be32_to_cpu(ctx->log_sz_usr_page) >> 24) & 0x1f);
+		break;
+	case CQ_LOG_PG_SZ:
+		param = (ctx->log_pg_sz & 0x1f) + 12;
+		break;
+	}
+
+out:
+	kfree(out);
+	return param;
+}
+
+static ssize_t dbg_read(struct file *filp, char __user *buf, size_t count,
+			loff_t *pos)
+{
+	struct mlx5_field_desc *desc;
+	struct mlx5_rsc_debug *d;
+	u64 field;
+	int ret;
+	int err;
+	char tbuf[18];
+
+	if (*pos)
+		return 0;
+
+	desc = filp->private_data;
+	d = (void *)(desc - desc->i) - sizeof(*d);
+	switch (d->type) {
+	case MLX5_DBG_RSC_QP:
+		field = qp_read_field(d->dev, d->object, desc->i);
+		break;
+
+	case MLX5_DBG_RSC_EQ:
+		field = eq_read_field(d->dev, d->object, desc->i);
+		break;
+
+	case MLX5_DBG_RSC_CQ:
+		field = cq_read_field(d->dev, d->object, desc->i);
+		break;
+
+	default:
+		mlx5_core_warn(d->dev, "invalid resource type %d\n", d->type);
+		return -EINVAL;
+	}
+
+	ret = snprintf(tbuf, sizeof(tbuf), "0x%llx\n", field);
+	if (ret > 0) {
+		err = copy_to_user(buf, tbuf, ret);
+		if (err)
+			return err;
+	}
+
+	*pos += ret;
+	return ret;
+}
+
+static const struct file_operations fops = {
+	.owner	= THIS_MODULE,
+	.open	= dbg_open,
+	.read	= dbg_read,
+};
+
+static int add_res_tree(struct mlx5_core_dev *dev, enum dbg_rsc_type type,
+			struct dentry *root, struct mlx5_rsc_debug **dbg,
+			int rsn, char **field, int nfile, void *data)
+{
+	char resn[32];
+	int i;
+	struct mlx5_rsc_debug *d;
+	int err;
+
+	d = kzalloc(sizeof(*d) + nfile * sizeof(d->fields[0]), GFP_KERNEL);
+	if (!d)
+		return -ENOMEM;
+
+	d->dev = dev;
+	d->object = data;
+	d->type = type;
+	sprintf(resn, "0x%x", rsn);
+	d->root = debugfs_create_dir(resn,  root);
+	if (!d->root) {
+		err = -ENOMEM;
+		goto out_free;
+	}
+
+	for (i = 0; i < nfile; ++i) {
+		d->fields[i].i = i;
+		d->fields[i].dent = debugfs_create_file(field[i], 0400,
+							d->root, &d->fields[i],
+							&fops);
+		if (!d->fields[i].dent) {
+			err = -ENOMEM;
+			goto out_rem;
+		}
+	}
+	*dbg = d;
+
+	return 0;
+out_rem:
+	debugfs_remove_recursive(d->root);
+
+out_free:
+	kfree(d);
+	return err;
+}
+
+static void rem_res_tree(struct mlx5_rsc_debug *d)
+{
+	debugfs_remove_recursive(d->root);
+	kfree(d);
+}
+
+int mlx5_debug_qp_add(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_QP, dev->priv.qp_debugfs,
+			   &qp->dbg, qp->qpn, qp_fields,
+			   ARRAY_SIZE(qp_fields), qp);
+	if (err)
+		qp->dbg = NULL;
+
+	return err;
+}
+
+void mlx5_debug_qp_remove(struct mlx5_core_dev *dev, struct mlx5_core_qp *qp)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (qp->dbg)
+		rem_res_tree(qp->dbg);
+}
+
+
+int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_EQ, dev->priv.eq_debugfs,
+			   &eq->dbg, eq->eqn, eq_fields,
+			   ARRAY_SIZE(eq_fields), eq);
+	if (err)
+		eq->dbg = NULL;
+
+	return err;
+}
+
+void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (eq->dbg)
+		rem_res_tree(eq->dbg);
+}
+
+int mlx5_debug_cq_add(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	int err;
+
+	if (!mlx5_debugfs_root)
+		return 0;
+
+	err = add_res_tree(dev, MLX5_DBG_RSC_CQ, dev->priv.cq_debugfs,
+			   &cq->dbg, cq->cqn, cq_fields,
+			   ARRAY_SIZE(cq_fields), cq);
+	if (err)
+		cq->dbg = NULL;
+
+	return err;
+}
+
+void mlx5_debug_cq_remove(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq)
+{
+	if (!mlx5_debugfs_root)
+		return;
+
+	if (cq->dbg)
+		rem_res_tree(cq->dbg);
+}