Merge branch 'cxgb4-hw-debug-logs'

Rahul Lakkireddy says:

====================
cxgb4: add support to get hardware debug logs via ethtool

This series of patches add support to collect hardware debug logs
via ethtool --get-dump facility.

Currently supports:

Memory dumps - Collects on-chip EDC0 and EDC1 dumps.
Hardware dumps - Collects firmware and hardware dumps.

Patch 1 adds ethtool set/get dump data.  It also adds template header
that precedes dump data.  This template header gives additional
information needed for extracting and decoding the collected dump
data.

Patch 2 adds base to collect dumps.  Also collects regdump.

Patch 3 collects on-chip EDC0 and EDC1 memory dumps.

Patch 4 collects firmware mbox log and device log.

Patch 5 updates base API for accessing TP indirect registers.

Patch 6 collects hardware TP module dump.

Patch 7 collects hardware SGE, PCIE, PM, UP CIM, MA, and HMA
module dumps.

Patch 8 collects hardware IBQ and OBQ dump.

Thanks,
Rahul

---
v2:
- Prefix symbols that pollute global namespace in files starting
  with cxgb4_* with "cxgb4_"
- Prefix symbols that pollute global namespace in files starting
  with cudbg_* with "cudbg_"
- Make cudbg_collect_mem_info() static.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

15 files changed, 2021 insertions, 142 deletions

diff --git a/drivers/net/ethernet/chelsio/cxgb4/Makefile b/drivers/net/ethernet/chelsio/cxgb4/Makefile
index fecd7aab673b..70d454379996 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/Makefile
+++ b/drivers/net/ethernet/chelsio/cxgb4/Makefile
@@ -6,7 +6,8 @@ obj-$(CONFIG_CHELSIO_T4) += cxgb4.o
 
 cxgb4-objs := cxgb4_main.o l2t.o t4_hw.o sge.o clip_tbl.o cxgb4_ethtool.o \
 	      cxgb4_uld.o sched.o cxgb4_filter.o cxgb4_tc_u32.o \
-	      cxgb4_ptp.o cxgb4_tc_flower.o
+	      cxgb4_ptp.o cxgb4_tc_flower.o cxgb4_cudbg.o \
+	      cudbg_common.o cudbg_lib.o
 cxgb4-$(CONFIG_CHELSIO_T4_DCB) +=  cxgb4_dcb.o
 cxgb4-$(CONFIG_CHELSIO_T4_FCOE) +=  cxgb4_fcoe.o
 cxgb4-$(CONFIG_DEBUG_FS) += cxgb4_debugfs.o
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_common.c b/drivers/net/ethernet/chelsio/cxgb4/cudbg_common.c
new file mode 100644
index 000000000000..f78ba1743b5a
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_common.c
@@ -0,0 +1,54 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#include "cxgb4.h"
+#include "cudbg_if.h"
+#include "cudbg_lib_common.h"
+
+int cudbg_get_buff(struct cudbg_buffer *pdbg_buff, u32 size,
+		   struct cudbg_buffer *pin_buff)
+{
+	u32 offset;
+
+	offset = pdbg_buff->offset;
+	if (offset + size > pdbg_buff->size)
+		return CUDBG_STATUS_NO_MEM;
+
+	pin_buff->data = (char *)pdbg_buff->data + offset;
+	pin_buff->offset = offset;
+	pin_buff->size = size;
+	pdbg_buff->size -= size;
+	return 0;
+}
+
+void cudbg_put_buff(struct cudbg_buffer *pin_buff,
+		    struct cudbg_buffer *pdbg_buff)
+{
+	pdbg_buff->size += pin_buff->size;
+	pin_buff->data = NULL;
+	pin_buff->offset = 0;
+	pin_buff->size = 0;
+}
+
+void cudbg_update_buff(struct cudbg_buffer *pin_buff,
+		       struct cudbg_buffer *pout_buff)
+{
+	/* We already write to buffer provided by ethool, so just
+	 * increment offset to next free space.
+	 */
+	pout_buff->offset += pin_buff->size;
+}
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h
new file mode 100644
index 000000000000..d7f3392f618f
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_entity.h
@@ -0,0 +1,181 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#ifndef __CUDBG_ENTITY_H__
+#define __CUDBG_ENTITY_H__
+
+#define EDC0_FLAG 3
+#define EDC1_FLAG 4
+
+struct card_mem {
+	u16 size_edc0;
+	u16 size_edc1;
+	u16 mem_flag;
+};
+
+struct cudbg_mbox_log {
+	struct mbox_cmd entry;
+	u32 hi[MBOX_LEN / 8];
+	u32 lo[MBOX_LEN / 8];
+};
+
+struct ireg_field {
+	u32 ireg_addr;
+	u32 ireg_data;
+	u32 ireg_local_offset;
+	u32 ireg_offset_range;
+};
+
+struct ireg_buf {
+	struct ireg_field tp_pio;
+	u32 outbuf[32];
+};
+
+#define IREG_NUM_ELEM 4
+
+static const u32 t6_tp_pio_array[][IREG_NUM_ELEM] = {
+	{0x7e40, 0x7e44, 0x020, 28}, /* t6_tp_pio_regs_20_to_3b */
+	{0x7e40, 0x7e44, 0x040, 10}, /* t6_tp_pio_regs_40_to_49 */
+	{0x7e40, 0x7e44, 0x050, 10}, /* t6_tp_pio_regs_50_to_59 */
+	{0x7e40, 0x7e44, 0x060, 14}, /* t6_tp_pio_regs_60_to_6d */
+	{0x7e40, 0x7e44, 0x06F, 1}, /* t6_tp_pio_regs_6f */
+	{0x7e40, 0x7e44, 0x070, 6}, /* t6_tp_pio_regs_70_to_75 */
+	{0x7e40, 0x7e44, 0x130, 18}, /* t6_tp_pio_regs_130_to_141 */
+	{0x7e40, 0x7e44, 0x145, 19}, /* t6_tp_pio_regs_145_to_157 */
+	{0x7e40, 0x7e44, 0x160, 1}, /* t6_tp_pio_regs_160 */
+	{0x7e40, 0x7e44, 0x230, 25}, /* t6_tp_pio_regs_230_to_248 */
+	{0x7e40, 0x7e44, 0x24a, 3}, /* t6_tp_pio_regs_24c */
+	{0x7e40, 0x7e44, 0x8C0, 1} /* t6_tp_pio_regs_8c0 */
+};
+
+static const u32 t5_tp_pio_array[][IREG_NUM_ELEM] = {
+	{0x7e40, 0x7e44, 0x020, 28}, /* t5_tp_pio_regs_20_to_3b */
+	{0x7e40, 0x7e44, 0x040, 19}, /* t5_tp_pio_regs_40_to_52 */
+	{0x7e40, 0x7e44, 0x054, 2}, /* t5_tp_pio_regs_54_to_55 */
+	{0x7e40, 0x7e44, 0x060, 13}, /* t5_tp_pio_regs_60_to_6c */
+	{0x7e40, 0x7e44, 0x06F, 1}, /* t5_tp_pio_regs_6f */
+	{0x7e40, 0x7e44, 0x120, 4}, /* t5_tp_pio_regs_120_to_123 */
+	{0x7e40, 0x7e44, 0x12b, 2}, /* t5_tp_pio_regs_12b_to_12c */
+	{0x7e40, 0x7e44, 0x12f, 21}, /* t5_tp_pio_regs_12f_to_143 */
+	{0x7e40, 0x7e44, 0x145, 19}, /* t5_tp_pio_regs_145_to_157 */
+	{0x7e40, 0x7e44, 0x230, 25}, /* t5_tp_pio_regs_230_to_248 */
+	{0x7e40, 0x7e44, 0x8C0, 1} /* t5_tp_pio_regs_8c0 */
+};
+
+static const u32 t6_tp_tm_pio_array[][IREG_NUM_ELEM] = {
+	{0x7e18, 0x7e1c, 0x0, 12}
+};
+
+static const u32 t5_tp_tm_pio_array[][IREG_NUM_ELEM] = {
+	{0x7e18, 0x7e1c, 0x0, 12}
+};
+
+static const u32 t6_tp_mib_index_array[6][IREG_NUM_ELEM] = {
+	{0x7e50, 0x7e54, 0x0, 13},
+	{0x7e50, 0x7e54, 0x10, 6},
+	{0x7e50, 0x7e54, 0x18, 21},
+	{0x7e50, 0x7e54, 0x30, 32},
+	{0x7e50, 0x7e54, 0x50, 22},
+	{0x7e50, 0x7e54, 0x68, 12}
+};
+
+static const u32 t5_tp_mib_index_array[9][IREG_NUM_ELEM] = {
+	{0x7e50, 0x7e54, 0x0, 13},
+	{0x7e50, 0x7e54, 0x10, 6},
+	{0x7e50, 0x7e54, 0x18, 8},
+	{0x7e50, 0x7e54, 0x20, 13},
+	{0x7e50, 0x7e54, 0x30, 16},
+	{0x7e50, 0x7e54, 0x40, 16},
+	{0x7e50, 0x7e54, 0x50, 16},
+	{0x7e50, 0x7e54, 0x60, 6},
+	{0x7e50, 0x7e54, 0x68, 4}
+};
+
+static const u32 t5_sge_dbg_index_array[2][IREG_NUM_ELEM] = {
+	{0x10cc, 0x10d0, 0x0, 16},
+	{0x10cc, 0x10d4, 0x0, 16},
+};
+
+static const u32 t5_pcie_pdbg_array[][IREG_NUM_ELEM] = {
+	{0x5a04, 0x5a0c, 0x00, 0x20}, /* t5_pcie_pdbg_regs_00_to_20 */
+	{0x5a04, 0x5a0c, 0x21, 0x20}, /* t5_pcie_pdbg_regs_21_to_40 */
+	{0x5a04, 0x5a0c, 0x41, 0x10}, /* t5_pcie_pdbg_regs_41_to_50 */
+};
+
+static const u32 t5_pcie_cdbg_array[][IREG_NUM_ELEM] = {
+	{0x5a10, 0x5a18, 0x00, 0x20}, /* t5_pcie_cdbg_regs_00_to_20 */
+	{0x5a10, 0x5a18, 0x21, 0x18}, /* t5_pcie_cdbg_regs_21_to_37 */
+};
+
+static const u32 t5_pm_rx_array[][IREG_NUM_ELEM] = {
+	{0x8FD0, 0x8FD4, 0x10000, 0x20}, /* t5_pm_rx_regs_10000_to_10020 */
+	{0x8FD0, 0x8FD4, 0x10021, 0x0D}, /* t5_pm_rx_regs_10021_to_1002c */
+};
+
+static const u32 t5_pm_tx_array[][IREG_NUM_ELEM] = {
+	{0x8FF0, 0x8FF4, 0x10000, 0x20}, /* t5_pm_tx_regs_10000_to_10020 */
+	{0x8FF0, 0x8FF4, 0x10021, 0x1D}, /* t5_pm_tx_regs_10021_to_1003c */
+};
+
+static const u32 t6_ma_ireg_array[][IREG_NUM_ELEM] = {
+	{0x78f8, 0x78fc, 0xa000, 23}, /* t6_ma_regs_a000_to_a016 */
+	{0x78f8, 0x78fc, 0xa400, 30}, /* t6_ma_regs_a400_to_a41e */
+	{0x78f8, 0x78fc, 0xa800, 20} /* t6_ma_regs_a800_to_a813 */
+};
+
+static const u32 t6_ma_ireg_array2[][IREG_NUM_ELEM] = {
+	{0x78f8, 0x78fc, 0xe400, 17}, /* t6_ma_regs_e400_to_e600 */
+	{0x78f8, 0x78fc, 0xe640, 13} /* t6_ma_regs_e640_to_e7c0 */
+};
+
+static const u32 t6_up_cim_reg_array[][IREG_NUM_ELEM] = {
+	{0x7b50, 0x7b54, 0x2000, 0x20}, /* up_cim_2000_to_207c */
+	{0x7b50, 0x7b54, 0x2080, 0x1d}, /* up_cim_2080_to_20fc */
+	{0x7b50, 0x7b54, 0x00, 0x20}, /* up_cim_00_to_7c */
+	{0x7b50, 0x7b54, 0x80, 0x20}, /* up_cim_80_to_fc */
+	{0x7b50, 0x7b54, 0x100, 0x11}, /* up_cim_100_to_14c */
+	{0x7b50, 0x7b54, 0x200, 0x10}, /* up_cim_200_to_23c */
+	{0x7b50, 0x7b54, 0x240, 0x2}, /* up_cim_240_to_244 */
+	{0x7b50, 0x7b54, 0x250, 0x2}, /* up_cim_250_to_254 */
+	{0x7b50, 0x7b54, 0x260, 0x2}, /* up_cim_260_to_264 */
+	{0x7b50, 0x7b54, 0x270, 0x2}, /* up_cim_270_to_274 */
+	{0x7b50, 0x7b54, 0x280, 0x20}, /* up_cim_280_to_2fc */
+	{0x7b50, 0x7b54, 0x300, 0x20}, /* up_cim_300_to_37c */
+	{0x7b50, 0x7b54, 0x380, 0x14}, /* up_cim_380_to_3cc */
+
+};
+
+static const u32 t5_up_cim_reg_array[][IREG_NUM_ELEM] = {
+	{0x7b50, 0x7b54, 0x2000, 0x20}, /* up_cim_2000_to_207c */
+	{0x7b50, 0x7b54, 0x2080, 0x19}, /* up_cim_2080_to_20ec */
+	{0x7b50, 0x7b54, 0x00, 0x20}, /* up_cim_00_to_7c */
+	{0x7b50, 0x7b54, 0x80, 0x20}, /* up_cim_80_to_fc */
+	{0x7b50, 0x7b54, 0x100, 0x11}, /* up_cim_100_to_14c */
+	{0x7b50, 0x7b54, 0x200, 0x10}, /* up_cim_200_to_23c */
+	{0x7b50, 0x7b54, 0x240, 0x2}, /* up_cim_240_to_244 */
+	{0x7b50, 0x7b54, 0x250, 0x2}, /* up_cim_250_to_254 */
+	{0x7b50, 0x7b54, 0x260, 0x2}, /* up_cim_260_to_264 */
+	{0x7b50, 0x7b54, 0x270, 0x2}, /* up_cim_270_to_274 */
+	{0x7b50, 0x7b54, 0x280, 0x20}, /* up_cim_280_to_2fc */
+	{0x7b50, 0x7b54, 0x300, 0x20}, /* up_cim_300_to_37c */
+	{0x7b50, 0x7b54, 0x380, 0x14}, /* up_cim_380_to_3cc */
+};
+
+static const u32 t6_hma_ireg_array[][IREG_NUM_ELEM] = {
+	{0x51320, 0x51324, 0xa000, 32} /* t6_hma_regs_a000_to_a01f */
+};
+#endif /* __CUDBG_ENTITY_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h
new file mode 100644
index 000000000000..9b8005e67811
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_if.h
@@ -0,0 +1,69 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#ifndef __CUDBG_IF_H__
+#define __CUDBG_IF_H__
+
+/* Error codes */
+#define CUDBG_STATUS_NO_MEM -19
+#define CUDBG_STATUS_ENTITY_NOT_FOUND -24
+#define CUDBG_SYSTEM_ERROR -29
+
+#define CUDBG_MAJOR_VERSION 1
+#define CUDBG_MINOR_VERSION 14
+
+enum cudbg_dbg_entity_type {
+	CUDBG_REG_DUMP = 1,
+	CUDBG_DEV_LOG = 2,
+	CUDBG_CIM_IBQ_TP0 = 6,
+	CUDBG_CIM_IBQ_TP1 = 7,
+	CUDBG_CIM_IBQ_ULP = 8,
+	CUDBG_CIM_IBQ_SGE0 = 9,
+	CUDBG_CIM_IBQ_SGE1 = 10,
+	CUDBG_CIM_IBQ_NCSI = 11,
+	CUDBG_CIM_OBQ_ULP0 = 12,
+	CUDBG_CIM_OBQ_ULP1 = 13,
+	CUDBG_CIM_OBQ_ULP2 = 14,
+	CUDBG_CIM_OBQ_ULP3 = 15,
+	CUDBG_CIM_OBQ_SGE = 16,
+	CUDBG_CIM_OBQ_NCSI = 17,
+	CUDBG_EDC0 = 18,
+	CUDBG_EDC1 = 19,
+	CUDBG_TP_INDIRECT = 36,
+	CUDBG_SGE_INDIRECT = 37,
+	CUDBG_CIM_OBQ_RXQ0 = 47,
+	CUDBG_CIM_OBQ_RXQ1 = 48,
+	CUDBG_PCIE_INDIRECT = 50,
+	CUDBG_PM_INDIRECT = 51,
+	CUDBG_MA_INDIRECT = 61,
+	CUDBG_UP_CIM_INDIRECT = 64,
+	CUDBG_MBOX_LOG = 66,
+	CUDBG_HMA_INDIRECT = 67,
+	CUDBG_MAX_ENTITY = 70,
+};
+
+struct cudbg_init {
+	struct adapter *adap; /* Pointer to adapter structure */
+	void *outbuf; /* Output buffer */
+	u32 outbuf_size;  /* Output buffer size */
+};
+
+static inline unsigned int cudbg_mbytes_to_bytes(unsigned int size)
+{
+	return size * 1024 * 1024;
+}
+#endif /* __CUDBG_IF_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c
new file mode 100644
index 000000000000..c451b2e42a6c
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.c
@@ -0,0 +1,867 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#include "t4_regs.h"
+#include "cxgb4.h"
+#include "cudbg_if.h"
+#include "cudbg_lib_common.h"
+#include "cudbg_lib.h"
+#include "cudbg_entity.h"
+
+static void cudbg_write_and_release_buff(struct cudbg_buffer *pin_buff,
+					 struct cudbg_buffer *dbg_buff)
+{
+	cudbg_update_buff(pin_buff, dbg_buff);
+	cudbg_put_buff(pin_buff, dbg_buff);
+}
+
+static int is_fw_attached(struct cudbg_init *pdbg_init)
+{
+	struct adapter *padap = pdbg_init->adap;
+
+	if (!(padap->flags & FW_OK) || padap->use_bd)
+		return 0;
+
+	return 1;
+}
+
+/* This function will add additional padding bytes into debug_buffer to make it
+ * 4 byte aligned.
+ */
+void cudbg_align_debug_buffer(struct cudbg_buffer *dbg_buff,
+			      struct cudbg_entity_hdr *entity_hdr)
+{
+	u8 zero_buf[4] = {0};
+	u8 padding, remain;
+
+	remain = (dbg_buff->offset - entity_hdr->start_offset) % 4;
+	padding = 4 - remain;
+	if (remain) {
+		memcpy(((u8 *)dbg_buff->data) + dbg_buff->offset, &zero_buf,
+		       padding);
+		dbg_buff->offset += padding;
+		entity_hdr->num_pad = padding;
+	}
+	entity_hdr->size = dbg_buff->offset - entity_hdr->start_offset;
+}
+
+struct cudbg_entity_hdr *cudbg_get_entity_hdr(void *outbuf, int i)
+{
+	struct cudbg_hdr *cudbg_hdr = (struct cudbg_hdr *)outbuf;
+
+	return (struct cudbg_entity_hdr *)
+	       ((char *)outbuf + cudbg_hdr->hdr_len +
+		(sizeof(struct cudbg_entity_hdr) * (i - 1)));
+}
+
+int cudbg_collect_reg_dump(struct cudbg_init *pdbg_init,
+			   struct cudbg_buffer *dbg_buff,
+			   struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	u32 buf_size = 0;
+	int rc = 0;
+
+	if (is_t4(padap->params.chip))
+		buf_size = T4_REGMAP_SIZE;
+	else if (is_t5(padap->params.chip) || is_t6(padap->params.chip))
+		buf_size = T5_REGMAP_SIZE;
+
+	rc = cudbg_get_buff(dbg_buff, buf_size, &temp_buff);
+	if (rc)
+		return rc;
+	t4_get_regs(padap, (void *)temp_buff.data, temp_buff.size);
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_fw_devlog(struct cudbg_init *pdbg_init,
+			    struct cudbg_buffer *dbg_buff,
+			    struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct devlog_params *dparams;
+	int rc = 0;
+
+	rc = t4_init_devlog_params(padap);
+	if (rc < 0) {
+		cudbg_err->sys_err = rc;
+		return rc;
+	}
+
+	dparams = &padap->params.devlog;
+	rc = cudbg_get_buff(dbg_buff, dparams->size, &temp_buff);
+	if (rc)
+		return rc;
+
+	/* Collect FW devlog */
+	if (dparams->start != 0) {
+		spin_lock(&padap->win0_lock);
+		rc = t4_memory_rw(padap, padap->params.drv_memwin,
+				  dparams->memtype, dparams->start,
+				  dparams->size,
+				  (__be32 *)(char *)temp_buff.data,
+				  1);
+		spin_unlock(&padap->win0_lock);
+		if (rc) {
+			cudbg_err->sys_err = rc;
+			cudbg_put_buff(&temp_buff, dbg_buff);
+			return rc;
+		}
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+static int cudbg_read_cim_ibq(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err, int qid)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	int no_of_read_words, rc = 0;
+	u32 qsize;
+
+	/* collect CIM IBQ */
+	qsize = CIM_IBQ_SIZE * 4 * sizeof(u32);
+	rc = cudbg_get_buff(dbg_buff, qsize, &temp_buff);
+	if (rc)
+		return rc;
+
+	/* t4_read_cim_ibq will return no. of read words or error */
+	no_of_read_words = t4_read_cim_ibq(padap, qid,
+					   (u32 *)((u32 *)temp_buff.data +
+					   temp_buff.offset), qsize);
+	/* no_of_read_words is less than or equal to 0 means error */
+	if (no_of_read_words <= 0) {
+		if (!no_of_read_words)
+			rc = CUDBG_SYSTEM_ERROR;
+		else
+			rc = no_of_read_words;
+		cudbg_err->sys_err = rc;
+		cudbg_put_buff(&temp_buff, dbg_buff);
+		return rc;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_cim_ibq_tp0(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 0);
+}
+
+int cudbg_collect_cim_ibq_tp1(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 1);
+}
+
+int cudbg_collect_cim_ibq_ulp(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 2);
+}
+
+int cudbg_collect_cim_ibq_sge0(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 3);
+}
+
+int cudbg_collect_cim_ibq_sge1(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 4);
+}
+
+int cudbg_collect_cim_ibq_ncsi(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_ibq(pdbg_init, dbg_buff, cudbg_err, 5);
+}
+
+static int cudbg_read_cim_obq(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err, int qid)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	int no_of_read_words, rc = 0;
+	u32 qsize;
+
+	/* collect CIM OBQ */
+	qsize =  6 * CIM_OBQ_SIZE * 4 *  sizeof(u32);
+	rc = cudbg_get_buff(dbg_buff, qsize, &temp_buff);
+	if (rc)
+		return rc;
+
+	/* t4_read_cim_obq will return no. of read words or error */
+	no_of_read_words = t4_read_cim_obq(padap, qid,
+					   (u32 *)((u32 *)temp_buff.data +
+					   temp_buff.offset), qsize);
+	/* no_of_read_words is less than or equal to 0 means error */
+	if (no_of_read_words <= 0) {
+		if (!no_of_read_words)
+			rc = CUDBG_SYSTEM_ERROR;
+		else
+			rc = no_of_read_words;
+		cudbg_err->sys_err = rc;
+		cudbg_put_buff(&temp_buff, dbg_buff);
+		return rc;
+	}
+	temp_buff.size = no_of_read_words * 4;
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_cim_obq_ulp0(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 0);
+}
+
+int cudbg_collect_cim_obq_ulp1(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 1);
+}
+
+int cudbg_collect_cim_obq_ulp2(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 2);
+}
+
+int cudbg_collect_cim_obq_ulp3(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 3);
+}
+
+int cudbg_collect_cim_obq_sge(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 4);
+}
+
+int cudbg_collect_cim_obq_ncsi(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 5);
+}
+
+int cudbg_collect_obq_sge_rx_q0(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 6);
+}
+
+int cudbg_collect_obq_sge_rx_q1(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err)
+{
+	return cudbg_read_cim_obq(pdbg_init, dbg_buff, cudbg_err, 7);
+}
+
+static int cudbg_read_fw_mem(struct cudbg_init *pdbg_init,
+			     struct cudbg_buffer *dbg_buff, u8 mem_type,
+			     unsigned long tot_len,
+			     struct cudbg_error *cudbg_err)
+{
+	unsigned long bytes, bytes_left, bytes_read = 0;
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	int rc = 0;
+
+	bytes_left = tot_len;
+	while (bytes_left > 0) {
+		bytes = min_t(unsigned long, bytes_left,
+			      (unsigned long)CUDBG_CHUNK_SIZE);
+		rc = cudbg_get_buff(dbg_buff, bytes, &temp_buff);
+		if (rc)
+			return rc;
+		spin_lock(&padap->win0_lock);
+		rc = t4_memory_rw(padap, MEMWIN_NIC, mem_type,
+				  bytes_read, bytes,
+				  (__be32 *)temp_buff.data,
+				  1);
+		spin_unlock(&padap->win0_lock);
+		if (rc) {
+			cudbg_err->sys_err = rc;
+			cudbg_put_buff(&temp_buff, dbg_buff);
+			return rc;
+		}
+		bytes_left -= bytes;
+		bytes_read += bytes;
+		cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	}
+	return rc;
+}
+
+static void cudbg_collect_mem_info(struct cudbg_init *pdbg_init,
+				   struct card_mem *mem_info)
+{
+	struct adapter *padap = pdbg_init->adap;
+	u32 value;
+
+	value = t4_read_reg(padap, MA_EDRAM0_BAR_A);
+	value = EDRAM0_SIZE_G(value);
+	mem_info->size_edc0 = (u16)value;
+
+	value = t4_read_reg(padap, MA_EDRAM1_BAR_A);
+	value = EDRAM1_SIZE_G(value);
+	mem_info->size_edc1 = (u16)value;
+
+	value = t4_read_reg(padap, MA_TARGET_MEM_ENABLE_A);
+	if (value & EDRAM0_ENABLE_F)
+		mem_info->mem_flag |= (1 << EDC0_FLAG);
+	if (value & EDRAM1_ENABLE_F)
+		mem_info->mem_flag |= (1 << EDC1_FLAG);
+}
+
+static void cudbg_t4_fwcache(struct cudbg_init *pdbg_init,
+			     struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	int rc;
+
+	if (is_fw_attached(pdbg_init)) {
+		/* Flush uP dcache before reading edcX/mcX  */
+		rc = t4_fwcache(padap, FW_PARAM_DEV_FWCACHE_FLUSH);
+		if (rc)
+			cudbg_err->sys_warn = rc;
+	}
+}
+
+static int cudbg_collect_mem_region(struct cudbg_init *pdbg_init,
+				    struct cudbg_buffer *dbg_buff,
+				    struct cudbg_error *cudbg_err,
+				    u8 mem_type)
+{
+	struct card_mem mem_info = {0};
+	unsigned long flag, size;
+	int rc;
+
+	cudbg_t4_fwcache(pdbg_init, cudbg_err);
+	cudbg_collect_mem_info(pdbg_init, &mem_info);
+	switch (mem_type) {
+	case MEM_EDC0:
+		flag = (1 << EDC0_FLAG);
+		size = cudbg_mbytes_to_bytes(mem_info.size_edc0);
+		break;
+	case MEM_EDC1:
+		flag = (1 << EDC1_FLAG);
+		size = cudbg_mbytes_to_bytes(mem_info.size_edc1);
+		break;
+	default:
+		rc = CUDBG_STATUS_ENTITY_NOT_FOUND;
+		goto err;
+	}
+
+	if (mem_info.mem_flag & flag) {
+		rc = cudbg_read_fw_mem(pdbg_init, dbg_buff, mem_type,
+				       size, cudbg_err);
+		if (rc)
+			goto err;
+	} else {
+		rc = CUDBG_STATUS_ENTITY_NOT_FOUND;
+		goto err;
+	}
+err:
+	return rc;
+}
+
+int cudbg_collect_edc0_meminfo(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_collect_mem_region(pdbg_init, dbg_buff, cudbg_err,
+					MEM_EDC0);
+}
+
+int cudbg_collect_edc1_meminfo(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	return cudbg_collect_mem_region(pdbg_init, dbg_buff, cudbg_err,
+					MEM_EDC1);
+}
+
+int cudbg_collect_tp_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *ch_tp_pio;
+	int i, rc, n = 0;
+	u32 size;
+
+	if (is_t5(padap->params.chip))
+		n = sizeof(t5_tp_pio_array) +
+		    sizeof(t5_tp_tm_pio_array) +
+		    sizeof(t5_tp_mib_index_array);
+	else
+		n = sizeof(t6_tp_pio_array) +
+		    sizeof(t6_tp_tm_pio_array) +
+		    sizeof(t6_tp_mib_index_array);
+
+	n = n / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	ch_tp_pio = (struct ireg_buf *)temp_buff.data;
+
+	/* TP_PIO */
+	if (is_t5(padap->params.chip))
+		n = sizeof(t5_tp_pio_array) / (IREG_NUM_ELEM * sizeof(u32));
+	else if (is_t6(padap->params.chip))
+		n = sizeof(t6_tp_pio_array) / (IREG_NUM_ELEM * sizeof(u32));
+
+	for (i = 0; i < n; i++) {
+		struct ireg_field *tp_pio = &ch_tp_pio->tp_pio;
+		u32 *buff = ch_tp_pio->outbuf;
+
+		if (is_t5(padap->params.chip)) {
+			tp_pio->ireg_addr = t5_tp_pio_array[i][0];
+			tp_pio->ireg_data = t5_tp_pio_array[i][1];
+			tp_pio->ireg_local_offset = t5_tp_pio_array[i][2];
+			tp_pio->ireg_offset_range = t5_tp_pio_array[i][3];
+		} else if (is_t6(padap->params.chip)) {
+			tp_pio->ireg_addr = t6_tp_pio_array[i][0];
+			tp_pio->ireg_data = t6_tp_pio_array[i][1];
+			tp_pio->ireg_local_offset = t6_tp_pio_array[i][2];
+			tp_pio->ireg_offset_range = t6_tp_pio_array[i][3];
+		}
+		t4_tp_pio_read(padap, buff, tp_pio->ireg_offset_range,
+			       tp_pio->ireg_local_offset, true);
+		ch_tp_pio++;
+	}
+
+	/* TP_TM_PIO */
+	if (is_t5(padap->params.chip))
+		n = sizeof(t5_tp_tm_pio_array) / (IREG_NUM_ELEM * sizeof(u32));
+	else if (is_t6(padap->params.chip))
+		n = sizeof(t6_tp_tm_pio_array) / (IREG_NUM_ELEM * sizeof(u32));
+
+	for (i = 0; i < n; i++) {
+		struct ireg_field *tp_pio = &ch_tp_pio->tp_pio;
+		u32 *buff = ch_tp_pio->outbuf;
+
+		if (is_t5(padap->params.chip)) {
+			tp_pio->ireg_addr = t5_tp_tm_pio_array[i][0];
+			tp_pio->ireg_data = t5_tp_tm_pio_array[i][1];
+			tp_pio->ireg_local_offset = t5_tp_tm_pio_array[i][2];
+			tp_pio->ireg_offset_range = t5_tp_tm_pio_array[i][3];
+		} else if (is_t6(padap->params.chip)) {
+			tp_pio->ireg_addr = t6_tp_tm_pio_array[i][0];
+			tp_pio->ireg_data = t6_tp_tm_pio_array[i][1];
+			tp_pio->ireg_local_offset = t6_tp_tm_pio_array[i][2];
+			tp_pio->ireg_offset_range = t6_tp_tm_pio_array[i][3];
+		}
+		t4_tp_tm_pio_read(padap, buff, tp_pio->ireg_offset_range,
+				  tp_pio->ireg_local_offset, true);
+		ch_tp_pio++;
+	}
+
+	/* TP_MIB_INDEX */
+	if (is_t5(padap->params.chip))
+		n = sizeof(t5_tp_mib_index_array) /
+		    (IREG_NUM_ELEM * sizeof(u32));
+	else if (is_t6(padap->params.chip))
+		n = sizeof(t6_tp_mib_index_array) /
+		    (IREG_NUM_ELEM * sizeof(u32));
+
+	for (i = 0; i < n ; i++) {
+		struct ireg_field *tp_pio = &ch_tp_pio->tp_pio;
+		u32 *buff = ch_tp_pio->outbuf;
+
+		if (is_t5(padap->params.chip)) {
+			tp_pio->ireg_addr = t5_tp_mib_index_array[i][0];
+			tp_pio->ireg_data = t5_tp_mib_index_array[i][1];
+			tp_pio->ireg_local_offset =
+				t5_tp_mib_index_array[i][2];
+			tp_pio->ireg_offset_range =
+				t5_tp_mib_index_array[i][3];
+		} else if (is_t6(padap->params.chip)) {
+			tp_pio->ireg_addr = t6_tp_mib_index_array[i][0];
+			tp_pio->ireg_data = t6_tp_mib_index_array[i][1];
+			tp_pio->ireg_local_offset =
+				t6_tp_mib_index_array[i][2];
+			tp_pio->ireg_offset_range =
+				t6_tp_mib_index_array[i][3];
+		}
+		t4_tp_mib_read(padap, buff, tp_pio->ireg_offset_range,
+			       tp_pio->ireg_local_offset, true);
+		ch_tp_pio++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_sge_indirect(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *ch_sge_dbg;
+	int i, rc;
+
+	rc = cudbg_get_buff(dbg_buff, sizeof(*ch_sge_dbg) * 2, &temp_buff);
+	if (rc)
+		return rc;
+
+	ch_sge_dbg = (struct ireg_buf *)temp_buff.data;
+	for (i = 0; i < 2; i++) {
+		struct ireg_field *sge_pio = &ch_sge_dbg->tp_pio;
+		u32 *buff = ch_sge_dbg->outbuf;
+
+		sge_pio->ireg_addr = t5_sge_dbg_index_array[i][0];
+		sge_pio->ireg_data = t5_sge_dbg_index_array[i][1];
+		sge_pio->ireg_local_offset = t5_sge_dbg_index_array[i][2];
+		sge_pio->ireg_offset_range = t5_sge_dbg_index_array[i][3];
+		t4_read_indirect(padap,
+				 sge_pio->ireg_addr,
+				 sge_pio->ireg_data,
+				 buff,
+				 sge_pio->ireg_offset_range,
+				 sge_pio->ireg_local_offset);
+		ch_sge_dbg++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_pcie_indirect(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *ch_pcie;
+	int i, rc, n;
+	u32 size;
+
+	n = sizeof(t5_pcie_pdbg_array) / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n * 2;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	ch_pcie = (struct ireg_buf *)temp_buff.data;
+	/* PCIE_PDBG */
+	for (i = 0; i < n; i++) {
+		struct ireg_field *pcie_pio = &ch_pcie->tp_pio;
+		u32 *buff = ch_pcie->outbuf;
+
+		pcie_pio->ireg_addr = t5_pcie_pdbg_array[i][0];
+		pcie_pio->ireg_data = t5_pcie_pdbg_array[i][1];
+		pcie_pio->ireg_local_offset = t5_pcie_pdbg_array[i][2];
+		pcie_pio->ireg_offset_range = t5_pcie_pdbg_array[i][3];
+		t4_read_indirect(padap,
+				 pcie_pio->ireg_addr,
+				 pcie_pio->ireg_data,
+				 buff,
+				 pcie_pio->ireg_offset_range,
+				 pcie_pio->ireg_local_offset);
+		ch_pcie++;
+	}
+
+	/* PCIE_CDBG */
+	n = sizeof(t5_pcie_cdbg_array) / (IREG_NUM_ELEM * sizeof(u32));
+	for (i = 0; i < n; i++) {
+		struct ireg_field *pcie_pio = &ch_pcie->tp_pio;
+		u32 *buff = ch_pcie->outbuf;
+
+		pcie_pio->ireg_addr = t5_pcie_cdbg_array[i][0];
+		pcie_pio->ireg_data = t5_pcie_cdbg_array[i][1];
+		pcie_pio->ireg_local_offset = t5_pcie_cdbg_array[i][2];
+		pcie_pio->ireg_offset_range = t5_pcie_cdbg_array[i][3];
+		t4_read_indirect(padap,
+				 pcie_pio->ireg_addr,
+				 pcie_pio->ireg_data,
+				 buff,
+				 pcie_pio->ireg_offset_range,
+				 pcie_pio->ireg_local_offset);
+		ch_pcie++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_pm_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *ch_pm;
+	int i, rc, n;
+	u32 size;
+
+	n = sizeof(t5_pm_rx_array) / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n * 2;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	ch_pm = (struct ireg_buf *)temp_buff.data;
+	/* PM_RX */
+	for (i = 0; i < n; i++) {
+		struct ireg_field *pm_pio = &ch_pm->tp_pio;
+		u32 *buff = ch_pm->outbuf;
+
+		pm_pio->ireg_addr = t5_pm_rx_array[i][0];
+		pm_pio->ireg_data = t5_pm_rx_array[i][1];
+		pm_pio->ireg_local_offset = t5_pm_rx_array[i][2];
+		pm_pio->ireg_offset_range = t5_pm_rx_array[i][3];
+		t4_read_indirect(padap,
+				 pm_pio->ireg_addr,
+				 pm_pio->ireg_data,
+				 buff,
+				 pm_pio->ireg_offset_range,
+				 pm_pio->ireg_local_offset);
+		ch_pm++;
+	}
+
+	/* PM_TX */
+	n = sizeof(t5_pm_tx_array) / (IREG_NUM_ELEM * sizeof(u32));
+	for (i = 0; i < n; i++) {
+		struct ireg_field *pm_pio = &ch_pm->tp_pio;
+		u32 *buff = ch_pm->outbuf;
+
+		pm_pio->ireg_addr = t5_pm_tx_array[i][0];
+		pm_pio->ireg_data = t5_pm_tx_array[i][1];
+		pm_pio->ireg_local_offset = t5_pm_tx_array[i][2];
+		pm_pio->ireg_offset_range = t5_pm_tx_array[i][3];
+		t4_read_indirect(padap,
+				 pm_pio->ireg_addr,
+				 pm_pio->ireg_data,
+				 buff,
+				 pm_pio->ireg_offset_range,
+				 pm_pio->ireg_local_offset);
+		ch_pm++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_ma_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *ma_indr;
+	int i, rc, n;
+	u32 size, j;
+
+	if (CHELSIO_CHIP_VERSION(padap->params.chip) < CHELSIO_T6)
+		return CUDBG_STATUS_ENTITY_NOT_FOUND;
+
+	n = sizeof(t6_ma_ireg_array) / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n * 2;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	ma_indr = (struct ireg_buf *)temp_buff.data;
+	for (i = 0; i < n; i++) {
+		struct ireg_field *ma_fli = &ma_indr->tp_pio;
+		u32 *buff = ma_indr->outbuf;
+
+		ma_fli->ireg_addr = t6_ma_ireg_array[i][0];
+		ma_fli->ireg_data = t6_ma_ireg_array[i][1];
+		ma_fli->ireg_local_offset = t6_ma_ireg_array[i][2];
+		ma_fli->ireg_offset_range = t6_ma_ireg_array[i][3];
+		t4_read_indirect(padap, ma_fli->ireg_addr, ma_fli->ireg_data,
+				 buff, ma_fli->ireg_offset_range,
+				 ma_fli->ireg_local_offset);
+		ma_indr++;
+	}
+
+	n = sizeof(t6_ma_ireg_array2) / (IREG_NUM_ELEM * sizeof(u32));
+	for (i = 0; i < n; i++) {
+		struct ireg_field *ma_fli = &ma_indr->tp_pio;
+		u32 *buff = ma_indr->outbuf;
+
+		ma_fli->ireg_addr = t6_ma_ireg_array2[i][0];
+		ma_fli->ireg_data = t6_ma_ireg_array2[i][1];
+		ma_fli->ireg_local_offset = t6_ma_ireg_array2[i][2];
+		for (j = 0; j < t6_ma_ireg_array2[i][3]; j++) {
+			t4_read_indirect(padap, ma_fli->ireg_addr,
+					 ma_fli->ireg_data, buff, 1,
+					 ma_fli->ireg_local_offset);
+			buff++;
+			ma_fli->ireg_local_offset += 0x20;
+		}
+		ma_indr++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_up_cim_indirect(struct cudbg_init *pdbg_init,
+				  struct cudbg_buffer *dbg_buff,
+				  struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *up_cim;
+	int i, rc, n;
+	u32 size;
+
+	n = sizeof(t5_up_cim_reg_array) / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	up_cim = (struct ireg_buf *)temp_buff.data;
+	for (i = 0; i < n; i++) {
+		struct ireg_field *up_cim_reg = &up_cim->tp_pio;
+		u32 *buff = up_cim->outbuf;
+
+		if (is_t5(padap->params.chip)) {
+			up_cim_reg->ireg_addr = t5_up_cim_reg_array[i][0];
+			up_cim_reg->ireg_data = t5_up_cim_reg_array[i][1];
+			up_cim_reg->ireg_local_offset =
+						t5_up_cim_reg_array[i][2];
+			up_cim_reg->ireg_offset_range =
+						t5_up_cim_reg_array[i][3];
+		} else if (is_t6(padap->params.chip)) {
+			up_cim_reg->ireg_addr = t6_up_cim_reg_array[i][0];
+			up_cim_reg->ireg_data = t6_up_cim_reg_array[i][1];
+			up_cim_reg->ireg_local_offset =
+						t6_up_cim_reg_array[i][2];
+			up_cim_reg->ireg_offset_range =
+						t6_up_cim_reg_array[i][3];
+		}
+
+		rc = t4_cim_read(padap, up_cim_reg->ireg_local_offset,
+				 up_cim_reg->ireg_offset_range, buff);
+		if (rc) {
+			cudbg_put_buff(&temp_buff, dbg_buff);
+			return rc;
+		}
+		up_cim++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_mbox_log(struct cudbg_init *pdbg_init,
+			   struct cudbg_buffer *dbg_buff,
+			   struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_mbox_log *mboxlog = NULL;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct mbox_cmd_log *log = NULL;
+	struct mbox_cmd *entry;
+	unsigned int entry_idx;
+	u16 mbox_cmds;
+	int i, k, rc;
+	u64 flit;
+	u32 size;
+
+	log = padap->mbox_log;
+	mbox_cmds = padap->mbox_log->size;
+	size = sizeof(struct cudbg_mbox_log) * mbox_cmds;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	mboxlog = (struct cudbg_mbox_log *)temp_buff.data;
+	for (k = 0; k < mbox_cmds; k++) {
+		entry_idx = log->cursor + k;
+		if (entry_idx >= log->size)
+			entry_idx -= log->size;
+
+		entry = mbox_cmd_log_entry(log, entry_idx);
+		/* skip over unused entries */
+		if (entry->timestamp == 0)
+			continue;
+
+		memcpy(&mboxlog->entry, entry, sizeof(struct mbox_cmd));
+		for (i = 0; i < MBOX_LEN / 8; i++) {
+			flit = entry->cmd[i];
+			mboxlog->hi[i] = (u32)(flit >> 32);
+			mboxlog->lo[i] = (u32)flit;
+		}
+		mboxlog++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
+
+int cudbg_collect_hma_indirect(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err)
+{
+	struct adapter *padap = pdbg_init->adap;
+	struct cudbg_buffer temp_buff = { 0 };
+	struct ireg_buf *hma_indr;
+	int i, rc, n;
+	u32 size;
+
+	if (CHELSIO_CHIP_VERSION(padap->params.chip) < CHELSIO_T6)
+		return CUDBG_STATUS_ENTITY_NOT_FOUND;
+
+	n = sizeof(t6_hma_ireg_array) / (IREG_NUM_ELEM * sizeof(u32));
+	size = sizeof(struct ireg_buf) * n;
+	rc = cudbg_get_buff(dbg_buff, size, &temp_buff);
+	if (rc)
+		return rc;
+
+	hma_indr = (struct ireg_buf *)temp_buff.data;
+	for (i = 0; i < n; i++) {
+		struct ireg_field *hma_fli = &hma_indr->tp_pio;
+		u32 *buff = hma_indr->outbuf;
+
+		hma_fli->ireg_addr = t6_hma_ireg_array[i][0];
+		hma_fli->ireg_data = t6_hma_ireg_array[i][1];
+		hma_fli->ireg_local_offset = t6_hma_ireg_array[i][2];
+		hma_fli->ireg_offset_range = t6_hma_ireg_array[i][3];
+		t4_read_indirect(padap, hma_fli->ireg_addr, hma_fli->ireg_data,
+				 buff, hma_fli->ireg_offset_range,
+				 hma_fli->ireg_local_offset);
+		hma_indr++;
+	}
+	cudbg_write_and_release_buff(&temp_buff, dbg_buff);
+	return rc;
+}
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h
new file mode 100644
index 000000000000..c4440c1d0142
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib.h
@@ -0,0 +1,103 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#ifndef __CUDBG_LIB_H__
+#define __CUDBG_LIB_H__
+
+int cudbg_collect_reg_dump(struct cudbg_init *pdbg_init,
+			   struct cudbg_buffer *dbg_buff,
+			   struct cudbg_error *cudbg_err);
+int cudbg_collect_fw_devlog(struct cudbg_init *pdbg_init,
+			    struct cudbg_buffer *dbg_buff,
+			    struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_tp0(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_tp1(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_ulp(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_sge0(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_sge1(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_ibq_ncsi(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_ulp0(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_ulp1(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_ulp2(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_ulp3(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_sge(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_cim_obq_ncsi(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_edc0_meminfo(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_edc1_meminfo(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_tp_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_sge_indirect(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+int cudbg_collect_obq_sge_rx_q0(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err);
+int cudbg_collect_obq_sge_rx_q1(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err);
+int cudbg_collect_pcie_indirect(struct cudbg_init *pdbg_init,
+				struct cudbg_buffer *dbg_buff,
+				struct cudbg_error *cudbg_err);
+int cudbg_collect_pm_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_ma_indirect(struct cudbg_init *pdbg_init,
+			      struct cudbg_buffer *dbg_buff,
+			      struct cudbg_error *cudbg_err);
+int cudbg_collect_up_cim_indirect(struct cudbg_init *pdbg_init,
+				  struct cudbg_buffer *dbg_buff,
+				  struct cudbg_error *cudbg_err);
+int cudbg_collect_mbox_log(struct cudbg_init *pdbg_init,
+			   struct cudbg_buffer *dbg_buff,
+			   struct cudbg_error *cudbg_err);
+int cudbg_collect_hma_indirect(struct cudbg_init *pdbg_init,
+			       struct cudbg_buffer *dbg_buff,
+			       struct cudbg_error *cudbg_err);
+
+struct cudbg_entity_hdr *cudbg_get_entity_hdr(void *outbuf, int i);
+void cudbg_align_debug_buffer(struct cudbg_buffer *dbg_buff,
+			      struct cudbg_entity_hdr *entity_hdr);
+#endif /* __CUDBG_LIB_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib_common.h b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib_common.h
new file mode 100644
index 000000000000..b150c5d1f7c0
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cudbg_lib_common.h
@@ -0,0 +1,81 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#ifndef __CUDBG_LIB_COMMON_H__
+#define __CUDBG_LIB_COMMON_H__
+
+#define CUDBG_SIGNATURE 67856866 /* CUDB in ascii */
+
+enum cudbg_dump_type {
+	CUDBG_DUMP_TYPE_MINI = 1,
+};
+
+enum cudbg_compression_type {
+	CUDBG_COMPRESSION_NONE = 1,
+};
+
+struct cudbg_hdr {
+	u32 signature;
+	u32 hdr_len;
+	u16 major_ver;
+	u16 minor_ver;
+	u32 data_len;
+	u32 hdr_flags;
+	u16 max_entities;
+	u8 chip_ver;
+	u8 dump_type:3;
+	u8 reserved1:1;
+	u8 compress_type:4;
+	u32 reserved[8];
+};
+
+struct cudbg_entity_hdr {
+	u32 entity_type;
+	u32 start_offset;
+	u32 size;
+	int hdr_flags;
+	u32 sys_warn;
+	u32 sys_err;
+	u8 num_pad;
+	u8 flag;             /* bit 0 is used to indicate ext data */
+	u8 reserved1[2];
+	u32 next_ext_offset; /* pointer to next extended entity meta data */
+	u32 reserved[5];
+};
+
+struct cudbg_buffer {
+	u32 size;
+	u32 offset;
+	char *data;
+};
+
+struct cudbg_error {
+	int sys_err;
+	int sys_warn;
+	int app_err;
+};
+
+#define CDUMP_MAX_COMP_BUF_SIZE ((64 * 1024) - 1)
+#define CUDBG_CHUNK_SIZE ((CDUMP_MAX_COMP_BUF_SIZE / 1024) * 1024)
+
+int cudbg_get_buff(struct cudbg_buffer *pdbg_buff, u32 size,
+		   struct cudbg_buffer *pin_buff);
+void cudbg_put_buff(struct cudbg_buffer *pin_buff,
+		    struct cudbg_buffer *pdbg_buff);
+void cudbg_update_buff(struct cudbg_buffer *pin_buff,
+		       struct cudbg_buffer *pout_buff);
+#endif /* __CUDBG_LIB_COMMON_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 0db3ab6ad094..4eaca05ebd3a 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -909,6 +909,9 @@ struct adapter {
 	/* TC flower offload */
 	DECLARE_HASHTABLE(flower_anymatch_tbl, 9);
 	struct timer_list flower_stats_timer;
+
+	/* Ethtool Dump */
+	struct ethtool_dump eth_dump;
 };
 
 /* Support for "sched-class" command to allow a TX Scheduling Class to be
@@ -1456,7 +1459,7 @@ unsigned int qtimer_val(const struct adapter *adap,
 
 int t4_init_devlog_params(struct adapter *adapter);
 int t4_init_sge_params(struct adapter *adapter);
-int t4_init_tp_params(struct adapter *adap);
+int t4_init_tp_params(struct adapter *adap, bool sleep_ok);
 int t4_filter_field_shift(const struct adapter *adap, int filter_sel);
 int t4_init_rss_mode(struct adapter *adap, int mbox);
 int t4_init_portinfo(struct port_info *pi, int mbox,
@@ -1470,14 +1473,15 @@ int t4_config_glbl_rss(struct adapter *adapter, int mbox, unsigned int mode,
 int t4_config_vi_rss(struct adapter *adapter, int mbox, unsigned int viid,
 		     unsigned int flags, unsigned int defq);
 int t4_read_rss(struct adapter *adapter, u16 *entries);
-void t4_read_rss_key(struct adapter *adapter, u32 *key);
-void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx);
+void t4_read_rss_key(struct adapter *adapter, u32 *key, bool sleep_ok);
+void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx,
+		      bool sleep_ok);
 void t4_read_rss_pf_config(struct adapter *adapter, unsigned int index,
-			   u32 *valp);
+			   u32 *valp, bool sleep_ok);
 void t4_read_rss_vf_config(struct adapter *adapter, unsigned int index,
-			   u32 *vfl, u32 *vfh);
-u32 t4_read_rss_pf_map(struct adapter *adapter);
-u32 t4_read_rss_pf_mask(struct adapter *adapter);
+			   u32 *vfl, u32 *vfh, bool sleep_ok);
+u32 t4_read_rss_pf_map(struct adapter *adapter, bool sleep_ok);
+u32 t4_read_rss_pf_mask(struct adapter *adapter, bool sleep_ok);
 
 unsigned int t4_get_mps_bg_map(struct adapter *adapter, int pidx);
 unsigned int t4_get_tp_ch_map(struct adapter *adapter, int pidx);
@@ -1508,14 +1512,18 @@ void t4_read_cong_tbl(struct adapter *adap, u16 incr[NMTUS][NCCTRL_WIN]);
 void t4_tp_wr_bits_indirect(struct adapter *adap, unsigned int addr,
 			    unsigned int mask, unsigned int val);
 void t4_tp_read_la(struct adapter *adap, u64 *la_buf, unsigned int *wrptr);
-void t4_tp_get_err_stats(struct adapter *adap, struct tp_err_stats *st);
-void t4_tp_get_cpl_stats(struct adapter *adap, struct tp_cpl_stats *st);
-void t4_tp_get_rdma_stats(struct adapter *adap, struct tp_rdma_stats *st);
-void t4_get_usm_stats(struct adapter *adap, struct tp_usm_stats *st);
+void t4_tp_get_err_stats(struct adapter *adap, struct tp_err_stats *st,
+			 bool sleep_ok);
+void t4_tp_get_cpl_stats(struct adapter *adap, struct tp_cpl_stats *st,
+			 bool sleep_ok);
+void t4_tp_get_rdma_stats(struct adapter *adap, struct tp_rdma_stats *st,
+			  bool sleep_ok);
+void t4_get_usm_stats(struct adapter *adap, struct tp_usm_stats *st,
+		      bool sleep_ok);
 void t4_tp_get_tcp_stats(struct adapter *adap, struct tp_tcp_stats *v4,
-			 struct tp_tcp_stats *v6);
+			 struct tp_tcp_stats *v6, bool sleep_ok);
 void t4_get_fcoe_stats(struct adapter *adap, unsigned int idx,
-		       struct tp_fcoe_stats *st);
+		       struct tp_fcoe_stats *st, bool sleep_ok);
 void t4_load_mtus(struct adapter *adap, const unsigned short *mtus,
 		  const unsigned short *alpha, const unsigned short *beta);
 
@@ -1624,6 +1632,13 @@ void t4_idma_monitor(struct adapter *adapter,
 		     int hz, int ticks);
 int t4_set_vf_mac_acl(struct adapter *adapter, unsigned int vf,
 		      unsigned int naddr, u8 *addr);
+void t4_tp_pio_read(struct adapter *adap, u32 *buff, u32 nregs,
+		    u32 start_index, bool sleep_ok);
+void t4_tp_tm_pio_read(struct adapter *adap, u32 *buff, u32 nregs,
+		       u32 start_index, bool sleep_ok);
+void t4_tp_mib_read(struct adapter *adap, u32 *buff, u32 nregs,
+		    u32 start_index, bool sleep_ok);
+
 void t4_uld_mem_free(struct adapter *adap);
 int t4_uld_mem_alloc(struct adapter *adap);
 void t4_uld_clean_up(struct adapter *adap);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c
new file mode 100644
index 000000000000..9d97080a9d17
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.c
@@ -0,0 +1,296 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#include "t4_regs.h"
+#include "cxgb4.h"
+#include "cxgb4_cudbg.h"
+#include "cudbg_entity.h"
+
+static const struct cxgb4_collect_entity cxgb4_collect_mem_dump[] = {
+	{ CUDBG_EDC0, cudbg_collect_edc0_meminfo },
+	{ CUDBG_EDC1, cudbg_collect_edc1_meminfo },
+};
+
+static const struct cxgb4_collect_entity cxgb4_collect_hw_dump[] = {
+	{ CUDBG_MBOX_LOG, cudbg_collect_mbox_log },
+	{ CUDBG_DEV_LOG, cudbg_collect_fw_devlog },
+	{ CUDBG_REG_DUMP, cudbg_collect_reg_dump },
+	{ CUDBG_CIM_IBQ_TP0, cudbg_collect_cim_ibq_tp0 },
+	{ CUDBG_CIM_IBQ_TP1, cudbg_collect_cim_ibq_tp1 },
+	{ CUDBG_CIM_IBQ_ULP, cudbg_collect_cim_ibq_ulp },
+	{ CUDBG_CIM_IBQ_SGE0, cudbg_collect_cim_ibq_sge0 },
+	{ CUDBG_CIM_IBQ_SGE1, cudbg_collect_cim_ibq_sge1 },
+	{ CUDBG_CIM_IBQ_NCSI, cudbg_collect_cim_ibq_ncsi },
+	{ CUDBG_CIM_OBQ_ULP0, cudbg_collect_cim_obq_ulp0 },
+	{ CUDBG_CIM_OBQ_ULP1, cudbg_collect_cim_obq_ulp1 },
+	{ CUDBG_CIM_OBQ_ULP2, cudbg_collect_cim_obq_ulp2 },
+	{ CUDBG_CIM_OBQ_ULP3, cudbg_collect_cim_obq_ulp3 },
+	{ CUDBG_CIM_OBQ_SGE, cudbg_collect_cim_obq_sge },
+	{ CUDBG_CIM_OBQ_NCSI, cudbg_collect_cim_obq_ncsi },
+	{ CUDBG_TP_INDIRECT, cudbg_collect_tp_indirect },
+	{ CUDBG_SGE_INDIRECT, cudbg_collect_sge_indirect },
+	{ CUDBG_CIM_OBQ_RXQ0, cudbg_collect_obq_sge_rx_q0 },
+	{ CUDBG_CIM_OBQ_RXQ1, cudbg_collect_obq_sge_rx_q1 },
+	{ CUDBG_PCIE_INDIRECT, cudbg_collect_pcie_indirect },
+	{ CUDBG_PM_INDIRECT, cudbg_collect_pm_indirect },
+	{ CUDBG_MA_INDIRECT, cudbg_collect_ma_indirect },
+	{ CUDBG_UP_CIM_INDIRECT, cudbg_collect_up_cim_indirect },
+	{ CUDBG_HMA_INDIRECT, cudbg_collect_hma_indirect },
+};
+
+static u32 cxgb4_get_entity_length(struct adapter *adap, u32 entity)
+{
+	u32 value, n = 0, len = 0;
+
+	switch (entity) {
+	case CUDBG_REG_DUMP:
+		switch (CHELSIO_CHIP_VERSION(adap->params.chip)) {
+		case CHELSIO_T4:
+			len = T4_REGMAP_SIZE;
+			break;
+		case CHELSIO_T5:
+		case CHELSIO_T6:
+			len = T5_REGMAP_SIZE;
+			break;
+		default:
+			break;
+		}
+		break;
+	case CUDBG_DEV_LOG:
+		len = adap->params.devlog.size;
+		break;
+	case CUDBG_CIM_IBQ_TP0:
+	case CUDBG_CIM_IBQ_TP1:
+	case CUDBG_CIM_IBQ_ULP:
+	case CUDBG_CIM_IBQ_SGE0:
+	case CUDBG_CIM_IBQ_SGE1:
+	case CUDBG_CIM_IBQ_NCSI:
+		len = CIM_IBQ_SIZE * 4 * sizeof(u32);
+		break;
+	case CUDBG_CIM_OBQ_ULP0:
+	case CUDBG_CIM_OBQ_ULP1:
+	case CUDBG_CIM_OBQ_ULP2:
+	case CUDBG_CIM_OBQ_ULP3:
+	case CUDBG_CIM_OBQ_SGE:
+	case CUDBG_CIM_OBQ_NCSI:
+	case CUDBG_CIM_OBQ_RXQ0:
+	case CUDBG_CIM_OBQ_RXQ1:
+		len = 6 * CIM_OBQ_SIZE * 4 * sizeof(u32);
+		break;
+	case CUDBG_EDC0:
+		value = t4_read_reg(adap, MA_TARGET_MEM_ENABLE_A);
+		if (value & EDRAM0_ENABLE_F) {
+			value = t4_read_reg(adap, MA_EDRAM0_BAR_A);
+			len = EDRAM0_SIZE_G(value);
+		}
+		len = cudbg_mbytes_to_bytes(len);
+		break;
+	case CUDBG_EDC1:
+		value = t4_read_reg(adap, MA_TARGET_MEM_ENABLE_A);
+		if (value & EDRAM1_ENABLE_F) {
+			value = t4_read_reg(adap, MA_EDRAM1_BAR_A);
+			len = EDRAM1_SIZE_G(value);
+		}
+		len = cudbg_mbytes_to_bytes(len);
+		break;
+	case CUDBG_TP_INDIRECT:
+		switch (CHELSIO_CHIP_VERSION(adap->params.chip)) {
+		case CHELSIO_T5:
+			n = sizeof(t5_tp_pio_array) +
+			    sizeof(t5_tp_tm_pio_array) +
+			    sizeof(t5_tp_mib_index_array);
+			break;
+		case CHELSIO_T6:
+			n = sizeof(t6_tp_pio_array) +
+			    sizeof(t6_tp_tm_pio_array) +
+			    sizeof(t6_tp_mib_index_array);
+			break;
+		default:
+			break;
+		}
+		n = n / (IREG_NUM_ELEM * sizeof(u32));
+		len = sizeof(struct ireg_buf) * n;
+		break;
+	case CUDBG_SGE_INDIRECT:
+		len = sizeof(struct ireg_buf) * 2;
+		break;
+	case CUDBG_PCIE_INDIRECT:
+		n = sizeof(t5_pcie_pdbg_array) / (IREG_NUM_ELEM * sizeof(u32));
+		len = sizeof(struct ireg_buf) * n * 2;
+		break;
+	case CUDBG_PM_INDIRECT:
+		n = sizeof(t5_pm_rx_array) / (IREG_NUM_ELEM * sizeof(u32));
+		len = sizeof(struct ireg_buf) * n * 2;
+		break;
+	case CUDBG_MA_INDIRECT:
+		if (CHELSIO_CHIP_VERSION(adap->params.chip) > CHELSIO_T5) {
+			n = sizeof(t6_ma_ireg_array) /
+			    (IREG_NUM_ELEM * sizeof(u32));
+			len = sizeof(struct ireg_buf) * n * 2;
+		}
+		break;
+	case CUDBG_UP_CIM_INDIRECT:
+		n = sizeof(t5_up_cim_reg_array) / (IREG_NUM_ELEM * sizeof(u32));
+		len = sizeof(struct ireg_buf) * n;
+		break;
+	case CUDBG_MBOX_LOG:
+		len = sizeof(struct cudbg_mbox_log) * adap->mbox_log->size;
+		break;
+	case CUDBG_HMA_INDIRECT:
+		if (CHELSIO_CHIP_VERSION(adap->params.chip) > CHELSIO_T5) {
+			n = sizeof(t6_hma_ireg_array) /
+			    (IREG_NUM_ELEM * sizeof(u32));
+			len = sizeof(struct ireg_buf) * n;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return len;
+}
+
+u32 cxgb4_get_dump_length(struct adapter *adap, u32 flag)
+{
+	u32 i, entity;
+	u32 len = 0;
+
+	if (flag & CXGB4_ETH_DUMP_HW) {
+		for (i = 0; i < ARRAY_SIZE(cxgb4_collect_hw_dump); i++) {
+			entity = cxgb4_collect_hw_dump[i].entity;
+			len += cxgb4_get_entity_length(adap, entity);
+		}
+	}
+
+	if (flag & CXGB4_ETH_DUMP_MEM) {
+		for (i = 0; i < ARRAY_SIZE(cxgb4_collect_mem_dump); i++) {
+			entity = cxgb4_collect_mem_dump[i].entity;
+			len += cxgb4_get_entity_length(adap, entity);
+		}
+	}
+
+	return len;
+}
+
+static void cxgb4_cudbg_collect_entity(struct cudbg_init *pdbg_init,
+				       struct cudbg_buffer *dbg_buff,
+				       const struct cxgb4_collect_entity *e_arr,
+				       u32 arr_size, void *buf, u32 *tot_size)
+{
+	struct adapter *adap = pdbg_init->adap;
+	struct cudbg_error cudbg_err = { 0 };
+	struct cudbg_entity_hdr *entity_hdr;
+	u32 entity_size, i;
+	u32 total_size = 0;
+	int ret;
+
+	for (i = 0; i < arr_size; i++) {
+		const struct cxgb4_collect_entity *e = &e_arr[i];
+
+		/* Skip entities that won't fit in output buffer */
+		entity_size = cxgb4_get_entity_length(adap, e->entity);
+		if (entity_size >
+		    pdbg_init->outbuf_size - *tot_size - total_size)
+			continue;
+
+		entity_hdr = cudbg_get_entity_hdr(buf, e->entity);
+		entity_hdr->entity_type = e->entity;
+		entity_hdr->start_offset = dbg_buff->offset;
+		memset(&cudbg_err, 0, sizeof(struct cudbg_error));
+		ret = e->collect_cb(pdbg_init, dbg_buff, &cudbg_err);
+		if (ret) {
+			entity_hdr->size = 0;
+			dbg_buff->offset = entity_hdr->start_offset;
+		} else {
+			cudbg_align_debug_buffer(dbg_buff, entity_hdr);
+		}
+
+		/* Log error and continue with next entity */
+		if (cudbg_err.sys_err)
+			ret = CUDBG_SYSTEM_ERROR;
+
+		entity_hdr->hdr_flags = ret;
+		entity_hdr->sys_err = cudbg_err.sys_err;
+		entity_hdr->sys_warn = cudbg_err.sys_warn;
+		total_size += entity_hdr->size;
+	}
+
+	*tot_size += total_size;
+}
+
+int cxgb4_cudbg_collect(struct adapter *adap, void *buf, u32 *buf_size,
+			u32 flag)
+{
+	struct cudbg_init cudbg_init = { 0 };
+	struct cudbg_buffer dbg_buff = { 0 };
+	u32 size, min_size, total_size = 0;
+	struct cudbg_hdr *cudbg_hdr;
+
+	size = *buf_size;
+
+	cudbg_init.adap = adap;
+	cudbg_init.outbuf = buf;
+	cudbg_init.outbuf_size = size;
+
+	dbg_buff.data = buf;
+	dbg_buff.size = size;
+	dbg_buff.offset = 0;
+
+	cudbg_hdr = (struct cudbg_hdr *)buf;
+	cudbg_hdr->signature = CUDBG_SIGNATURE;
+	cudbg_hdr->hdr_len = sizeof(struct cudbg_hdr);
+	cudbg_hdr->major_ver = CUDBG_MAJOR_VERSION;
+	cudbg_hdr->minor_ver = CUDBG_MINOR_VERSION;
+	cudbg_hdr->max_entities = CUDBG_MAX_ENTITY;
+	cudbg_hdr->chip_ver = adap->params.chip;
+	cudbg_hdr->dump_type = CUDBG_DUMP_TYPE_MINI;
+	cudbg_hdr->compress_type = CUDBG_COMPRESSION_NONE;
+
+	min_size = sizeof(struct cudbg_hdr) +
+		   sizeof(struct cudbg_entity_hdr) *
+		   cudbg_hdr->max_entities;
+	if (size < min_size)
+		return -ENOMEM;
+
+	dbg_buff.offset += min_size;
+	total_size = dbg_buff.offset;
+
+	if (flag & CXGB4_ETH_DUMP_HW)
+		cxgb4_cudbg_collect_entity(&cudbg_init, &dbg_buff,
+					   cxgb4_collect_hw_dump,
+					   ARRAY_SIZE(cxgb4_collect_hw_dump),
+					   buf,
+					   &total_size);
+
+	if (flag & CXGB4_ETH_DUMP_MEM)
+		cxgb4_cudbg_collect_entity(&cudbg_init, &dbg_buff,
+					   cxgb4_collect_mem_dump,
+					   ARRAY_SIZE(cxgb4_collect_mem_dump),
+					   buf,
+					   &total_size);
+
+	cudbg_hdr->data_len = total_size;
+	*buf_size = total_size;
+	return 0;
+}
+
+void cxgb4_init_ethtool_dump(struct adapter *adapter)
+{
+	adapter->eth_dump.flag = CXGB4_ETH_DUMP_NONE;
+	adapter->eth_dump.version = adapter->params.fw_vers;
+	adapter->eth_dump.len = 0;
+}
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.h
new file mode 100644
index 000000000000..c099b5aa2214
--- /dev/null
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_cudbg.h
@@ -0,0 +1,44 @@
+/*
+ *  Copyright (C) 2017 Chelsio Communications.  All rights reserved.
+ *
+ *  This program is free software; you can redistribute it and/or modify it
+ *  under the terms and conditions of the GNU General Public License,
+ *  version 2, as published by the Free Software Foundation.
+ *
+ *  This program is distributed in the hope it will be useful, but WITHOUT
+ *  ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ *  FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ *  more details.
+ *
+ *  The full GNU General Public License is included in this distribution in
+ *  the file called "COPYING".
+ *
+ */
+
+#ifndef __CXGB4_CUDBG_H__
+#define __CXGB4_CUDBG_H__
+
+#include "cudbg_if.h"
+#include "cudbg_lib_common.h"
+#include "cudbg_lib.h"
+
+typedef int (*cudbg_collect_callback_t)(struct cudbg_init *pdbg_init,
+					struct cudbg_buffer *dbg_buff,
+					struct cudbg_error *cudbg_err);
+
+struct cxgb4_collect_entity {
+	enum cudbg_dbg_entity_type entity;
+	cudbg_collect_callback_t collect_cb;
+};
+
+enum CXGB4_ETHTOOL_DUMP_FLAGS {
+	CXGB4_ETH_DUMP_NONE = ETH_FW_DUMP_DISABLE,
+	CXGB4_ETH_DUMP_MEM = (1 << 0), /* On-Chip Memory Dumps */
+	CXGB4_ETH_DUMP_HW = (1 << 1), /* various FW and HW dumps */
+};
+
+u32 cxgb4_get_dump_length(struct adapter *adap, u32 flag);
+int cxgb4_cudbg_collect(struct adapter *adap, void *buf, u32 *buf_size,
+			u32 flag);
+void cxgb4_init_ethtool_dump(struct adapter *adapter);
+#endif /* __CXGB4_CUDBG_H__ */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
index 76540b0e082d..917663b35603 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_debugfs.c
@@ -2211,7 +2211,7 @@ static int rss_key_show(struct seq_file *seq, void *v)
 {
 	u32 key[10];
 
-	t4_read_rss_key(seq->private, key);
+	t4_read_rss_key(seq->private, key, true);
 	seq_printf(seq, "%08x%08x%08x%08x%08x%08x%08x%08x%08x%08x\n",
 		   key[9], key[8], key[7], key[6], key[5], key[4], key[3],
 		   key[2], key[1], key[0]);
@@ -2248,7 +2248,7 @@ static ssize_t rss_key_write(struct file *file, const char __user *buf,
 		}
 	}
 
-	t4_write_rss_key(adap, key, -1);
+	t4_write_rss_key(adap, key, -1, true);
 	return count;
 }
 
@@ -2325,12 +2325,13 @@ static int rss_pf_config_open(struct inode *inode, struct file *file)
 		return -ENOMEM;
 
 	pfconf = (struct rss_pf_conf *)p->data;
-	rss_pf_map = t4_read_rss_pf_map(adapter);
-	rss_pf_mask = t4_read_rss_pf_mask(adapter);
+	rss_pf_map = t4_read_rss_pf_map(adapter, true);
+	rss_pf_mask = t4_read_rss_pf_mask(adapter, true);
 	for (pf = 0; pf < 8; pf++) {
 		pfconf[pf].rss_pf_map = rss_pf_map;
 		pfconf[pf].rss_pf_mask = rss_pf_mask;
-		t4_read_rss_pf_config(adapter, pf, &pfconf[pf].rss_pf_config);
+		t4_read_rss_pf_config(adapter, pf, &pfconf[pf].rss_pf_config,
+				      true);
 	}
 	return 0;
 }
@@ -2393,7 +2394,7 @@ static int rss_vf_config_open(struct inode *inode, struct file *file)
 	vfconf = (struct rss_vf_conf *)p->data;
 	for (vf = 0; vf < vfcount; vf++) {
 		t4_read_rss_vf_config(adapter, vf, &vfconf[vf].rss_vf_vfl,
-				      &vfconf[vf].rss_vf_vfh);
+				      &vfconf[vf].rss_vf_vfh, true);
 	}
 	return 0;
 }
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
index a71af1e587e2..1b7f6b9ccc8b 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_ethtool.c
@@ -21,6 +21,7 @@
 #include "cxgb4.h"
 #include "t4_regs.h"
 #include "t4fw_api.h"
+#include "cxgb4_cudbg.h"
 
 #define EEPROM_MAGIC 0x38E2F10C
 
@@ -335,10 +336,10 @@ static void collect_adapter_stats(struct adapter *adap, struct adapter_stats *s)
 	memset(s, 0, sizeof(*s));
 
 	spin_lock(&adap->stats_lock);
-	t4_tp_get_tcp_stats(adap, &v4, &v6);
-	t4_tp_get_rdma_stats(adap, &rdma_stats);
-	t4_get_usm_stats(adap, &usm_stats);
-	t4_tp_get_err_stats(adap, &err_stats);
+	t4_tp_get_tcp_stats(adap, &v4, &v6, false);
+	t4_tp_get_rdma_stats(adap, &rdma_stats, false);
+	t4_get_usm_stats(adap, &usm_stats, false);
+	t4_tp_get_err_stats(adap, &err_stats, false);
 	spin_unlock(&adap->stats_lock);
 
 	s->db_drop = adap->db_stats.db_drop;
@@ -388,9 +389,9 @@ static void collect_channel_stats(struct adapter *adap, struct channel_stats *s,
 	memset(s, 0, sizeof(*s));
 
 	spin_lock(&adap->stats_lock);
-	t4_tp_get_cpl_stats(adap, &cpl_stats);
-	t4_tp_get_err_stats(adap, &err_stats);
-	t4_get_fcoe_stats(adap, i, &fcoe_stats);
+	t4_tp_get_cpl_stats(adap, &cpl_stats, false);
+	t4_tp_get_err_stats(adap, &err_stats, false);
+	t4_get_fcoe_stats(adap, i, &fcoe_stats, false);
 	spin_unlock(&adap->stats_lock);
 
 	s->cpl_req = cpl_stats.req[i];
@@ -1374,6 +1375,56 @@ static int get_rxnfc(struct net_device *dev, struct ethtool_rxnfc *info,
 	return -EOPNOTSUPP;
 }
 
+static int set_dump(struct net_device *dev, struct ethtool_dump *eth_dump)
+{
+	struct adapter *adapter = netdev2adap(dev);
+	u32 len = 0;
+
+	len = sizeof(struct cudbg_hdr) +
+	      sizeof(struct cudbg_entity_hdr) * CUDBG_MAX_ENTITY;
+	len += cxgb4_get_dump_length(adapter, eth_dump->flag);
+
+	adapter->eth_dump.flag = eth_dump->flag;
+	adapter->eth_dump.len = len;
+	return 0;
+}
+
+static int get_dump_flag(struct net_device *dev, struct ethtool_dump *eth_dump)
+{
+	struct adapter *adapter = netdev2adap(dev);
+
+	eth_dump->flag = adapter->eth_dump.flag;
+	eth_dump->len = adapter->eth_dump.len;
+	eth_dump->version = adapter->eth_dump.version;
+	return 0;
+}
+
+static int get_dump_data(struct net_device *dev, struct ethtool_dump *eth_dump,
+			 void *buf)
+{
+	struct adapter *adapter = netdev2adap(dev);
+	u32 len = 0;
+	int ret = 0;
+
+	if (adapter->eth_dump.flag == CXGB4_ETH_DUMP_NONE)
+		return -ENOENT;
+
+	len = sizeof(struct cudbg_hdr) +
+	      sizeof(struct cudbg_entity_hdr) * CUDBG_MAX_ENTITY;
+	len += cxgb4_get_dump_length(adapter, adapter->eth_dump.flag);
+	if (eth_dump->len < len)
+		return -ENOMEM;
+
+	ret = cxgb4_cudbg_collect(adapter, buf, &len, adapter->eth_dump.flag);
+	if (ret)
+		return ret;
+
+	eth_dump->flag = adapter->eth_dump.flag;
+	eth_dump->len = len;
+	eth_dump->version = adapter->eth_dump.version;
+	return 0;
+}
+
 static const struct ethtool_ops cxgb_ethtool_ops = {
 	.get_link_ksettings = get_link_ksettings,
 	.set_link_ksettings = set_link_ksettings,
@@ -1404,7 +1455,10 @@ static const struct ethtool_ops cxgb_ethtool_ops = {
 	.get_rxfh	   = get_rss_table,
 	.set_rxfh	   = set_rss_table,
 	.flash_device      = set_flash,
-	.get_ts_info       = get_ts_info
+	.get_ts_info       = get_ts_info,
+	.set_dump          = set_dump,
+	.get_dump_flag     = get_dump_flag,
+	.get_dump_data     = get_dump_data,
 };
 
 void cxgb4_set_ethtool_ops(struct net_device *netdev)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index fe4cbe22d5d7..8d97ae6039aa 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -81,6 +81,7 @@
 #include "cxgb4_tc_u32.h"
 #include "cxgb4_tc_flower.h"
 #include "cxgb4_ptp.h"
+#include "cxgb4_cudbg.h"
 
 char cxgb4_driver_name[] = KBUILD_MODNAME;
 
@@ -1638,7 +1639,7 @@ void cxgb4_get_tcp_stats(struct pci_dev *pdev, struct tp_tcp_stats *v4,
 	struct adapter *adap = pci_get_drvdata(pdev);
 
 	spin_lock(&adap->stats_lock);
-	t4_tp_get_tcp_stats(adap, v4, v6);
+	t4_tp_get_tcp_stats(adap, v4, v6, false);
 	spin_unlock(&adap->stats_lock);
 }
 EXPORT_SYMBOL(cxgb4_get_tcp_stats);
@@ -4076,7 +4077,7 @@ static int adap_init0(struct adapter *adap)
 	}
 	t4_init_sge_params(adap);
 	adap->flags |= FW_OK;
-	t4_init_tp_params(adap);
+	t4_init_tp_params(adap, true);
 	return 0;
 
 	/*
@@ -5035,6 +5036,8 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		cxgb4_set_ethtool_ops(netdev);
 	}
 
+	cxgb4_init_ethtool_dump(adapter);
+
 	pci_set_drvdata(pdev, adapter);
 
 	if (adapter->flags & FW_OK) {
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index b3fd1f457639..8fa40f9e75c4 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -5052,23 +5052,26 @@ static unsigned int t4_use_ldst(struct adapter *adap)
 }
 
 /**
- *	t4_fw_tp_pio_rw - Access TP PIO through LDST
- *	@adap: the adapter
- *	@vals: where the indirect register values are stored/written
- *	@nregs: how many indirect registers to read/write
- *	@start_idx: index of first indirect register to read/write
- *	@rw: Read (1) or Write (0)
+ * t4_tp_fw_ldst_rw - Access TP indirect register through LDST
+ * @adap: the adapter
+ * @cmd: TP fw ldst address space type
+ * @vals: where the indirect register values are stored/written
+ * @nregs: how many indirect registers to read/write
+ * @start_idx: index of first indirect register to read/write
+ * @rw: Read (1) or Write (0)
+ * @sleep_ok: if true we may sleep while awaiting command completion
  *
- *	Access TP PIO registers through LDST
+ * Access TP indirect registers through LDST
  */
-static void t4_fw_tp_pio_rw(struct adapter *adap, u32 *vals, unsigned int nregs,
-			    unsigned int start_index, unsigned int rw)
+static int t4_tp_fw_ldst_rw(struct adapter *adap, int cmd, u32 *vals,
+			    unsigned int nregs, unsigned int start_index,
+			    unsigned int rw, bool sleep_ok)
 {
-	int ret, i;
-	int cmd = FW_LDST_ADDRSPC_TP_PIO;
+	int ret = 0;
+	unsigned int i;
 	struct fw_ldst_cmd c;
 
-	for (i = 0 ; i < nregs; i++) {
+	for (i = 0; i < nregs; i++) {
 		memset(&c, 0, sizeof(c));
 		c.op_to_addrspace = cpu_to_be32(FW_CMD_OP_V(FW_LDST_CMD) |
 						FW_CMD_REQUEST_F |
@@ -5079,26 +5082,147 @@ static void t4_fw_tp_pio_rw(struct adapter *adap, u32 *vals, unsigned int nregs,
 
 		c.u.addrval.addr = cpu_to_be32(start_index + i);
 		c.u.addrval.val  = rw ? 0 : cpu_to_be32(vals[i]);
-		ret = t4_wr_mbox(adap, adap->mbox, &c, sizeof(c), &c);
-		if (!ret && rw)
+		ret = t4_wr_mbox_meat(adap, adap->mbox, &c, sizeof(c), &c,
+				      sleep_ok);
+		if (ret)
+			return ret;
+
+		if (rw)
 			vals[i] = be32_to_cpu(c.u.addrval.val);
 	}
+	return 0;
+}
+
+/**
+ * t4_tp_indirect_rw - Read/Write TP indirect register through LDST or backdoor
+ * @adap: the adapter
+ * @reg_addr: Address Register
+ * @reg_data: Data register
+ * @buff: where the indirect register values are stored/written
+ * @nregs: how many indirect registers to read/write
+ * @start_index: index of first indirect register to read/write
+ * @rw: READ(1) or WRITE(0)
+ * @sleep_ok: if true we may sleep while awaiting command completion
+ *
+ * Read/Write TP indirect registers through LDST if possible.
+ * Else, use backdoor access
+ **/
+static void t4_tp_indirect_rw(struct adapter *adap, u32 reg_addr, u32 reg_data,
+			      u32 *buff, u32 nregs, u32 start_index, int rw,
+			      bool sleep_ok)
+{
+	int rc = -EINVAL;
+	int cmd;
+
+	switch (reg_addr) {
+	case TP_PIO_ADDR_A:
+		cmd = FW_LDST_ADDRSPC_TP_PIO;
+		break;
+	case TP_TM_PIO_ADDR_A:
+		cmd = FW_LDST_ADDRSPC_TP_TM_PIO;
+		break;
+	case TP_MIB_INDEX_A:
+		cmd = FW_LDST_ADDRSPC_TP_MIB;
+		break;
+	default:
+		goto indirect_access;
+	}
+
+	if (t4_use_ldst(adap))
+		rc = t4_tp_fw_ldst_rw(adap, cmd, buff, nregs, start_index, rw,
+				      sleep_ok);
+
+indirect_access:
+
+	if (rc) {
+		if (rw)
+			t4_read_indirect(adap, reg_addr, reg_data, buff, nregs,
+					 start_index);
+		else
+			t4_write_indirect(adap, reg_addr, reg_data, buff, nregs,
+					  start_index);
+	}
+}
+
+/**
+ * t4_tp_pio_read - Read TP PIO registers
+ * @adap: the adapter
+ * @buff: where the indirect register values are written
+ * @nregs: how many indirect registers to read
+ * @start_index: index of first indirect register to read
+ * @sleep_ok: if true we may sleep while awaiting command completion
+ *
+ * Read TP PIO Registers
+ **/
+void t4_tp_pio_read(struct adapter *adap, u32 *buff, u32 nregs,
+		    u32 start_index, bool sleep_ok)
+{
+	t4_tp_indirect_rw(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A, buff, nregs,
+			  start_index, 1, sleep_ok);
+}
+
+/**
+ * t4_tp_pio_write - Write TP PIO registers
+ * @adap: the adapter
+ * @buff: where the indirect register values are stored
+ * @nregs: how many indirect registers to write
+ * @start_index: index of first indirect register to write
+ * @sleep_ok: if true we may sleep while awaiting command completion
+ *
+ * Write TP PIO Registers
+ **/
+static void t4_tp_pio_write(struct adapter *adap, u32 *buff, u32 nregs,
+			    u32 start_index, bool sleep_ok)
+{
+	t4_tp_indirect_rw(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A, buff, nregs,
+			  start_index, 0, sleep_ok);
+}
+
+/**
+ * t4_tp_tm_pio_read - Read TP TM PIO registers
+ * @adap: the adapter
+ * @buff: where the indirect register values are written
+ * @nregs: how many indirect registers to read
+ * @start_index: index of first indirect register to read
+ * @sleep_ok: if true we may sleep while awaiting command completion
+ *
+ * Read TP TM PIO Registers
+ **/
+void t4_tp_tm_pio_read(struct adapter *adap, u32 *buff, u32 nregs,
+		       u32 start_index, bool sleep_ok)
+{
+	t4_tp_indirect_rw(adap, TP_TM_PIO_ADDR_A, TP_TM_PIO_DATA_A, buff,
+			  nregs, start_index, 1, sleep_ok);
+}
+
+/**
+ * t4_tp_mib_read - Read TP MIB registers
+ * @adap: the adapter
+ * @buff: where the indirect register values are written
+ * @nregs: how many indirect registers to read
+ * @start_index: index of first indirect register to read
+ * @sleep_ok: if true we may sleep while awaiting command completion
+ *
+ * Read TP MIB Registers
+ **/
+void t4_tp_mib_read(struct adapter *adap, u32 *buff, u32 nregs, u32 start_index,
+		    bool sleep_ok)
+{
+	t4_tp_indirect_rw(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, buff, nregs,
+			  start_index, 1, sleep_ok);
 }
 
 /**
  *	t4_read_rss_key - read the global RSS key
  *	@adap: the adapter
  *	@key: 10-entry array holding the 320-bit RSS key
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Reads the global 320-bit RSS key.
  */
-void t4_read_rss_key(struct adapter *adap, u32 *key)
+void t4_read_rss_key(struct adapter *adap, u32 *key, bool sleep_ok)
 {
-	if (t4_use_ldst(adap))
-		t4_fw_tp_pio_rw(adap, key, 10, TP_RSS_SECRET_KEY0_A, 1);
-	else
-		t4_read_indirect(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A, key, 10,
-				 TP_RSS_SECRET_KEY0_A);
+	t4_tp_pio_read(adap, key, 10, TP_RSS_SECRET_KEY0_A, sleep_ok);
 }
 
 /**
@@ -5106,12 +5230,14 @@ void t4_read_rss_key(struct adapter *adap, u32 *key)
  *	@adap: the adapter
  *	@key: 10-entry array holding the 320-bit RSS key
  *	@idx: which RSS key to write
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Writes one of the RSS keys with the given 320-bit value.  If @idx is
  *	0..15 the corresponding entry in the RSS key table is written,
  *	otherwise the global RSS key is written.
  */
-void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx)
+void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx,
+		      bool sleep_ok)
 {
 	u8 rss_key_addr_cnt = 16;
 	u32 vrt = t4_read_reg(adap, TP_RSS_CONFIG_VRT_A);
@@ -5124,11 +5250,7 @@ void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx)
 	    (vrt & KEYEXTEND_F) && (KEYMODE_G(vrt) == 3))
 		rss_key_addr_cnt = 32;
 
-	if (t4_use_ldst(adap))
-		t4_fw_tp_pio_rw(adap, (void *)key, 10, TP_RSS_SECRET_KEY0_A, 0);
-	else
-		t4_write_indirect(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A, key, 10,
-				  TP_RSS_SECRET_KEY0_A);
+	t4_tp_pio_write(adap, (void *)key, 10, TP_RSS_SECRET_KEY0_A, sleep_ok);
 
 	if (idx >= 0 && idx < rss_key_addr_cnt) {
 		if (rss_key_addr_cnt > 16)
@@ -5146,19 +5268,15 @@ void t4_write_rss_key(struct adapter *adap, const u32 *key, int idx)
  *	@adapter: the adapter
  *	@index: the entry in the PF RSS table to read
  *	@valp: where to store the returned value
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Reads the PF RSS Configuration Table at the specified index and returns
  *	the value found there.
  */
 void t4_read_rss_pf_config(struct adapter *adapter, unsigned int index,
-			   u32 *valp)
+			   u32 *valp, bool sleep_ok)
 {
-	if (t4_use_ldst(adapter))
-		t4_fw_tp_pio_rw(adapter, valp, 1,
-				TP_RSS_PF0_CONFIG_A + index, 1);
-	else
-		t4_read_indirect(adapter, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 valp, 1, TP_RSS_PF0_CONFIG_A + index);
+	t4_tp_pio_read(adapter, valp, 1, TP_RSS_PF0_CONFIG_A + index, sleep_ok);
 }
 
 /**
@@ -5167,12 +5285,13 @@ void t4_read_rss_pf_config(struct adapter *adapter, unsigned int index,
  *	@index: the entry in the VF RSS table to read
  *	@vfl: where to store the returned VFL
  *	@vfh: where to store the returned VFH
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Reads the VF RSS Configuration Table at the specified index and returns
  *	the (VFL, VFH) values found there.
  */
 void t4_read_rss_vf_config(struct adapter *adapter, unsigned int index,
-			   u32 *vfl, u32 *vfh)
+			   u32 *vfl, u32 *vfh, bool sleep_ok)
 {
 	u32 vrt, mask, data;
 
@@ -5193,50 +5312,37 @@ void t4_read_rss_vf_config(struct adapter *adapter, unsigned int index,
 
 	/* Grab the VFL/VFH values ...
 	 */
-	if (t4_use_ldst(adapter)) {
-		t4_fw_tp_pio_rw(adapter, vfl, 1, TP_RSS_VFL_CONFIG_A, 1);
-		t4_fw_tp_pio_rw(adapter, vfh, 1, TP_RSS_VFH_CONFIG_A, 1);
-	} else {
-		t4_read_indirect(adapter, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 vfl, 1, TP_RSS_VFL_CONFIG_A);
-		t4_read_indirect(adapter, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 vfh, 1, TP_RSS_VFH_CONFIG_A);
-	}
+	t4_tp_pio_read(adapter, vfl, 1, TP_RSS_VFL_CONFIG_A, sleep_ok);
+	t4_tp_pio_read(adapter, vfh, 1, TP_RSS_VFH_CONFIG_A, sleep_ok);
 }
 
 /**
  *	t4_read_rss_pf_map - read PF RSS Map
  *	@adapter: the adapter
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Reads the PF RSS Map register and returns its value.
  */
-u32 t4_read_rss_pf_map(struct adapter *adapter)
+u32 t4_read_rss_pf_map(struct adapter *adapter, bool sleep_ok)
 {
 	u32 pfmap;
 
-	if (t4_use_ldst(adapter))
-		t4_fw_tp_pio_rw(adapter, &pfmap, 1, TP_RSS_PF_MAP_A, 1);
-	else
-		t4_read_indirect(adapter, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 &pfmap, 1, TP_RSS_PF_MAP_A);
+	t4_tp_pio_read(adapter, &pfmap, 1, TP_RSS_PF_MAP_A, sleep_ok);
 	return pfmap;
 }
 
 /**
  *	t4_read_rss_pf_mask - read PF RSS Mask
  *	@adapter: the adapter
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Reads the PF RSS Mask register and returns its value.
  */
-u32 t4_read_rss_pf_mask(struct adapter *adapter)
+u32 t4_read_rss_pf_mask(struct adapter *adapter, bool sleep_ok)
 {
 	u32 pfmask;
 
-	if (t4_use_ldst(adapter))
-		t4_fw_tp_pio_rw(adapter, &pfmask, 1, TP_RSS_PF_MSK_A, 1);
-	else
-		t4_read_indirect(adapter, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 &pfmask, 1, TP_RSS_PF_MSK_A);
+	t4_tp_pio_read(adapter, &pfmask, 1, TP_RSS_PF_MSK_A, sleep_ok);
 	return pfmask;
 }
 
@@ -5245,12 +5351,13 @@ u32 t4_read_rss_pf_mask(struct adapter *adapter)
  *	@adap: the adapter
  *	@v4: holds the TCP/IP counter values
  *	@v6: holds the TCP/IPv6 counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's TCP/IP and TCP/IPv6 MIB counters.
  *	Either @v4 or @v6 may be %NULL to skip the corresponding stats.
  */
 void t4_tp_get_tcp_stats(struct adapter *adap, struct tp_tcp_stats *v4,
-			 struct tp_tcp_stats *v6)
+			 struct tp_tcp_stats *v6, bool sleep_ok)
 {
 	u32 val[TP_MIB_TCP_RXT_SEG_LO_A - TP_MIB_TCP_OUT_RST_A + 1];
 
@@ -5259,16 +5366,16 @@ void t4_tp_get_tcp_stats(struct adapter *adap, struct tp_tcp_stats *v4,
 #define STAT64(x)   (((u64)STAT(x##_HI) << 32) | STAT(x##_LO))
 
 	if (v4) {
-		t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, val,
-				 ARRAY_SIZE(val), TP_MIB_TCP_OUT_RST_A);
+		t4_tp_mib_read(adap, val, ARRAY_SIZE(val),
+			       TP_MIB_TCP_OUT_RST_A, sleep_ok);
 		v4->tcp_out_rsts = STAT(OUT_RST);
 		v4->tcp_in_segs  = STAT64(IN_SEG);
 		v4->tcp_out_segs = STAT64(OUT_SEG);
 		v4->tcp_retrans_segs = STAT64(RXT_SEG);
 	}
 	if (v6) {
-		t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, val,
-				 ARRAY_SIZE(val), TP_MIB_TCP_V6OUT_RST_A);
+		t4_tp_mib_read(adap, val, ARRAY_SIZE(val),
+			       TP_MIB_TCP_V6OUT_RST_A, sleep_ok);
 		v6->tcp_out_rsts = STAT(OUT_RST);
 		v6->tcp_in_segs  = STAT64(IN_SEG);
 		v6->tcp_out_segs = STAT64(OUT_SEG);
@@ -5283,63 +5390,66 @@ void t4_tp_get_tcp_stats(struct adapter *adap, struct tp_tcp_stats *v4,
  *	t4_tp_get_err_stats - read TP's error MIB counters
  *	@adap: the adapter
  *	@st: holds the counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's error counters.
  */
-void t4_tp_get_err_stats(struct adapter *adap, struct tp_err_stats *st)
+void t4_tp_get_err_stats(struct adapter *adap, struct tp_err_stats *st,
+			 bool sleep_ok)
 {
 	int nchan = adap->params.arch.nchan;
 
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->mac_in_errs, nchan, TP_MIB_MAC_IN_ERR_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->hdr_in_errs, nchan, TP_MIB_HDR_IN_ERR_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->tcp_in_errs, nchan, TP_MIB_TCP_IN_ERR_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->tnl_cong_drops, nchan, TP_MIB_TNL_CNG_DROP_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->ofld_chan_drops, nchan, TP_MIB_OFD_CHN_DROP_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->tnl_tx_drops, nchan, TP_MIB_TNL_DROP_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->ofld_vlan_drops, nchan, TP_MIB_OFD_VLN_DROP_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 st->tcp6_in_errs, nchan, TP_MIB_TCP_V6IN_ERR_0_A);
-
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A,
-			 &st->ofld_no_neigh, 2, TP_MIB_OFD_ARP_DROP_A);
+	t4_tp_mib_read(adap, st->mac_in_errs, nchan, TP_MIB_MAC_IN_ERR_0_A,
+		       sleep_ok);
+	t4_tp_mib_read(adap, st->hdr_in_errs, nchan, TP_MIB_HDR_IN_ERR_0_A,
+		       sleep_ok);
+	t4_tp_mib_read(adap, st->tcp_in_errs, nchan, TP_MIB_TCP_IN_ERR_0_A,
+		       sleep_ok);
+	t4_tp_mib_read(adap, st->tnl_cong_drops, nchan,
+		       TP_MIB_TNL_CNG_DROP_0_A, sleep_ok);
+	t4_tp_mib_read(adap, st->ofld_chan_drops, nchan,
+		       TP_MIB_OFD_CHN_DROP_0_A, sleep_ok);
+	t4_tp_mib_read(adap, st->tnl_tx_drops, nchan, TP_MIB_TNL_DROP_0_A,
+		       sleep_ok);
+	t4_tp_mib_read(adap, st->ofld_vlan_drops, nchan,
+		       TP_MIB_OFD_VLN_DROP_0_A, sleep_ok);
+	t4_tp_mib_read(adap, st->tcp6_in_errs, nchan,
+		       TP_MIB_TCP_V6IN_ERR_0_A, sleep_ok);
+	t4_tp_mib_read(adap, &st->ofld_no_neigh, 2, TP_MIB_OFD_ARP_DROP_A,
+		       sleep_ok);
 }
 
 /**
  *	t4_tp_get_cpl_stats - read TP's CPL MIB counters
  *	@adap: the adapter
  *	@st: holds the counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's CPL counters.
  */
-void t4_tp_get_cpl_stats(struct adapter *adap, struct tp_cpl_stats *st)
+void t4_tp_get_cpl_stats(struct adapter *adap, struct tp_cpl_stats *st,
+			 bool sleep_ok)
 {
 	int nchan = adap->params.arch.nchan;
 
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, st->req,
-			 nchan, TP_MIB_CPL_IN_REQ_0_A);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, st->rsp,
-			 nchan, TP_MIB_CPL_OUT_RSP_0_A);
+	t4_tp_mib_read(adap, st->req, nchan, TP_MIB_CPL_IN_REQ_0_A, sleep_ok);
 
+	t4_tp_mib_read(adap, st->rsp, nchan, TP_MIB_CPL_OUT_RSP_0_A, sleep_ok);
 }
 
 /**
  *	t4_tp_get_rdma_stats - read TP's RDMA MIB counters
  *	@adap: the adapter
  *	@st: holds the counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's RDMA counters.
  */
-void t4_tp_get_rdma_stats(struct adapter *adap, struct tp_rdma_stats *st)
+void t4_tp_get_rdma_stats(struct adapter *adap, struct tp_rdma_stats *st,
+			  bool sleep_ok)
 {
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, &st->rqe_dfr_pkt,
-			 2, TP_MIB_RQE_DFR_PKT_A);
+	t4_tp_mib_read(adap, &st->rqe_dfr_pkt, 2, TP_MIB_RQE_DFR_PKT_A,
+		       sleep_ok);
 }
 
 /**
@@ -5347,20 +5457,24 @@ void t4_tp_get_rdma_stats(struct adapter *adap, struct tp_rdma_stats *st)
  *	@adap: the adapter
  *	@idx: the port index
  *	@st: holds the counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's FCoE counters for the selected port.
  */
 void t4_get_fcoe_stats(struct adapter *adap, unsigned int idx,
-		       struct tp_fcoe_stats *st)
+		       struct tp_fcoe_stats *st, bool sleep_ok)
 {
 	u32 val[2];
 
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, &st->frames_ddp,
-			 1, TP_MIB_FCOE_DDP_0_A + idx);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, &st->frames_drop,
-			 1, TP_MIB_FCOE_DROP_0_A + idx);
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, val,
-			 2, TP_MIB_FCOE_BYTE_0_HI_A + 2 * idx);
+	t4_tp_mib_read(adap, &st->frames_ddp, 1, TP_MIB_FCOE_DDP_0_A + idx,
+		       sleep_ok);
+
+	t4_tp_mib_read(adap, &st->frames_drop, 1,
+		       TP_MIB_FCOE_DROP_0_A + idx, sleep_ok);
+
+	t4_tp_mib_read(adap, val, 2, TP_MIB_FCOE_BYTE_0_HI_A + 2 * idx,
+		       sleep_ok);
+
 	st->octets_ddp = ((u64)val[0] << 32) | val[1];
 }
 
@@ -5368,15 +5482,16 @@ void t4_get_fcoe_stats(struct adapter *adap, unsigned int idx,
  *	t4_get_usm_stats - read TP's non-TCP DDP MIB counters
  *	@adap: the adapter
  *	@st: holds the counter values
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *	Returns the values of TP's counters for non-TCP directly-placed packets.
  */
-void t4_get_usm_stats(struct adapter *adap, struct tp_usm_stats *st)
+void t4_get_usm_stats(struct adapter *adap, struct tp_usm_stats *st,
+		      bool sleep_ok)
 {
 	u32 val[4];
 
-	t4_read_indirect(adap, TP_MIB_INDEX_A, TP_MIB_DATA_A, val, 4,
-			 TP_MIB_USM_PKTS_A);
+	t4_tp_mib_read(adap, val, 4, TP_MIB_USM_PKTS_A, sleep_ok);
 	st->frames = val[0];
 	st->drops = val[1];
 	st->octets = ((u64)val[2] << 32) | val[3];
@@ -8663,10 +8778,11 @@ int t4_init_sge_params(struct adapter *adapter)
 /**
  *      t4_init_tp_params - initialize adap->params.tp
  *      @adap: the adapter
+ *      @sleep_ok: if true we may sleep while awaiting command completion
  *
  *      Initialize various fields of the adapter's TP Parameters structure.
  */
-int t4_init_tp_params(struct adapter *adap)
+int t4_init_tp_params(struct adapter *adap, bool sleep_ok)
 {
 	int chan;
 	u32 v;
@@ -8682,19 +8798,11 @@ int t4_init_tp_params(struct adapter *adap)
 	/* Cache the adapter's Compressed Filter Mode and global Incress
 	 * Configuration.
 	 */
-	if (t4_use_ldst(adap)) {
-		t4_fw_tp_pio_rw(adap, &adap->params.tp.vlan_pri_map, 1,
-				TP_VLAN_PRI_MAP_A, 1);
-		t4_fw_tp_pio_rw(adap, &adap->params.tp.ingress_config, 1,
-				TP_INGRESS_CONFIG_A, 1);
-	} else {
-		t4_read_indirect(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 &adap->params.tp.vlan_pri_map, 1,
-				 TP_VLAN_PRI_MAP_A);
-		t4_read_indirect(adap, TP_PIO_ADDR_A, TP_PIO_DATA_A,
-				 &adap->params.tp.ingress_config, 1,
-				 TP_INGRESS_CONFIG_A);
-	}
+	t4_tp_pio_read(adap, &adap->params.tp.vlan_pri_map, 1,
+		       TP_VLAN_PRI_MAP_A, sleep_ok);
+	t4_tp_pio_read(adap, &adap->params.tp.ingress_config, 1,
+		       TP_INGRESS_CONFIG_A, sleep_ok);
+
 	/* For T6, cache the adapter's compressed error vector
 	 * and passing outer header info for encapsulated packets.
 	 */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
index dac90837842b..82614e078f50 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
@@ -1447,6 +1447,8 @@
 #define LKPTBLQUEUE0_M    0x3ffU
 #define LKPTBLQUEUE0_G(x) (((x) >> LKPTBLQUEUE0_S) & LKPTBLQUEUE0_M)
 
+#define TP_TM_PIO_ADDR_A 0x7e18
+#define TP_TM_PIO_DATA_A 0x7e1c
 #define TP_PIO_ADDR_A	0x7e40
 #define TP_PIO_DATA_A	0x7e44
 #define TP_MIB_INDEX_A	0x7e50