diff --git a/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.c b/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.c
new file mode 100644
index 0000000..7faa2f0
--- /dev/null
+++ b/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.c
@@ -0,0 +1,304 @@
+/*
+ * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <assert.h>
+#include <bpmp_ipc.h>
+#include <debug.h>
+#include <delay_timer.h>
+#include <errno.h>
+#include <mmio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <tegra_def.h>
+#include <utils_def.h>
+
+#include "intf.h"
+#include "ivc.h"
+
+/**
+ * Holds IVC channel data
+ */
+struct ccplex_bpmp_channel_data {
+	/* Buffer for incoming data */
+	struct frame_data *ib;
+
+	/* Buffer for outgoing data */
+	struct frame_data *ob;
+};
+
+static struct ccplex_bpmp_channel_data s_channel;
+static struct ivc ivc_ccplex_bpmp_channel;
+
+/*
+ * Helper functions to access the HSP doorbell registers
+ */
+static inline uint32_t hsp_db_read(uint32_t reg)
+{
+	return mmio_read_32((uint32_t)(TEGRA_HSP_DBELL_BASE + reg));
+}
+
+static inline void hsp_db_write(uint32_t reg, uint32_t val)
+{
+	mmio_write_32((uint32_t)(TEGRA_HSP_DBELL_BASE + reg), val);
+}
+
+/*******************************************************************************
+ *      IVC wrappers for CCPLEX <-> BPMP communication.
+ ******************************************************************************/
+
+static void tegra_bpmp_ring_bpmp_doorbell(void);
+
+/*
+ * Get the next frame where data can be written.
+ */
+static struct frame_data *tegra_bpmp_get_next_out_frame(void)
+{
+	struct frame_data *frame;
+	const struct ivc *ch = &ivc_ccplex_bpmp_channel;
+
+	frame = (struct frame_data *)tegra_ivc_write_get_next_frame(ch);
+	if (frame == NULL) {
+		ERROR("%s: Error in getting next frame, exiting\n", __func__);
+	} else {
+		s_channel.ob = frame;
+	}
+
+	return frame;
+}
+
+static void tegra_bpmp_signal_slave(void)
+{
+	(void)tegra_ivc_write_advance(&ivc_ccplex_bpmp_channel);
+	tegra_bpmp_ring_bpmp_doorbell();
+}
+
+static int32_t tegra_bpmp_free_master(void)
+{
+	return tegra_ivc_read_advance(&ivc_ccplex_bpmp_channel);
+}
+
+static bool tegra_bpmp_slave_acked(void)
+{
+	struct frame_data *frame;
+	bool ret = true;
+
+	frame = (struct frame_data *)tegra_ivc_read_get_next_frame(&ivc_ccplex_bpmp_channel);
+	if (frame == NULL) {
+		ret = false;
+	} else {
+		s_channel.ib = frame;
+	}
+
+	return ret;
+}
+
+static struct frame_data *tegra_bpmp_get_cur_in_frame(void)
+{
+	return s_channel.ib;
+}
+
+/*
+ * Enables BPMP to ring CCPlex doorbell
+ */
+static void tegra_bpmp_enable_ccplex_doorbell(void)
+{
+	uint32_t reg;
+
+	reg = hsp_db_read(HSP_DBELL_1_ENABLE);
+	reg |= HSP_MASTER_BPMP_BIT;
+	hsp_db_write(HSP_DBELL_1_ENABLE, reg);
+}
+
+/*
+ * CCPlex rings the BPMP doorbell
+ */
+static void tegra_bpmp_ring_bpmp_doorbell(void)
+{
+	/*
+	 * Any writes to this register has the same effect,
+	 * uses master ID of the write transaction and set
+	 * corresponding flag.
+	 */
+	hsp_db_write(HSP_DBELL_3_TRIGGER, HSP_MASTER_CCPLEX_BIT);
+}
+
+/*
+ * Returns true if CCPLex can ring BPMP doorbell, otherwise false.
+ * This also signals that BPMP is up and ready.
+ */
+static bool tegra_bpmp_can_ccplex_ring_doorbell(void)
+{
+	uint32_t reg;
+
+	/* check if ccplex can communicate with bpmp */
+	reg = hsp_db_read(HSP_DBELL_3_ENABLE);
+
+	return ((reg & HSP_MASTER_CCPLEX_BIT) != 0U);
+}
+
+static int32_t tegra_bpmp_wait_for_slave_ack(void)
+{
+	uint32_t timeout = TIMEOUT_RESPONSE_FROM_BPMP_US;
+
+	while (!tegra_bpmp_slave_acked() && (timeout != 0U)) {
+		udelay(1);
+		timeout--;
+	};
+
+	return ((timeout == 0U) ? -ETIMEDOUT : 0);
+}
+
+/*
+ * Notification from the ivc layer
+ */
+static void tegra_bpmp_ivc_notify(const struct ivc *ivc)
+{
+	(void)(ivc);
+
+	tegra_bpmp_ring_bpmp_doorbell();
+}
+
+/*
+ * Atomic send/receive API, which means it waits until slave acks
+ */
+static int32_t tegra_bpmp_ipc_send_req_atomic(uint32_t mrq, void *p_out,
+			uint32_t size_out, void *p_in, uint32_t size_in)
+{
+	struct frame_data *frame = tegra_bpmp_get_next_out_frame();
+	const struct frame_data *f_in = NULL;
+	int32_t ret = 0;
+	void *p_fdata;
+
+	if ((p_out == NULL) || (size_out > IVC_DATA_SZ_BYTES) ||
+	    (frame == NULL)) {
+		ERROR("%s: invalid parameters, exiting\n", __func__);
+		ret = -EINVAL;
+	}
+
+	if (ret == 0) {
+
+		/* prepare the command frame */
+		frame->mrq = mrq;
+		frame->flags = FLAG_DO_ACK;
+		p_fdata = frame->data;
+		(void)memcpy(p_fdata, p_out, (size_t)size_out);
+
+		/* signal the slave */
+		tegra_bpmp_signal_slave();
+
+		/* wait for slave to ack */
+		ret = tegra_bpmp_wait_for_slave_ack();
+		if (ret != 0) {
+			ERROR("failed waiting for the slave to ack\n");
+		}
+
+		/* retrieve the response frame */
+		if ((size_in <= IVC_DATA_SZ_BYTES) && (p_in != NULL) &&
+		    (ret == 0)) {
+
+			f_in = tegra_bpmp_get_cur_in_frame();
+			if (f_in != NULL) {
+				ERROR("Failed to get next input frame!\n");
+			} else {
+				(void)memcpy(p_in, p_fdata, (size_t)size_in);
+			}
+		}
+
+		if (ret == 0) {
+			ret = tegra_bpmp_free_master();
+			if (ret != 0) {
+				ERROR("Failed to free master\n");
+			}
+		}
+	}
+
+	return ret;
+}
+
+/*
+ * Initializes the BPMP<--->CCPlex communication path.
+ */
+int32_t tegra_bpmp_ipc_init(void)
+{
+	size_t msg_size;
+	uint32_t frame_size, timeout;
+	int32_t error = 0;
+
+	/* allow bpmp to ring CCPLEX's doorbell */
+	tegra_bpmp_enable_ccplex_doorbell();
+
+	/* wait for BPMP to actually ring the doorbell */
+	timeout = TIMEOUT_RESPONSE_FROM_BPMP_US;
+	while ((timeout != 0U) && !tegra_bpmp_can_ccplex_ring_doorbell()) {
+		udelay(1); /* bpmp turn-around time */
+		timeout--;
+	}
+
+	if (timeout == 0U) {
+		ERROR("%s: BPMP firmware is not ready\n", __func__);
+		return -ENOTSUP;
+	}
+
+	INFO("%s: BPMP handshake completed\n", __func__);
+
+	msg_size = tegra_ivc_align(IVC_CMD_SZ_BYTES);
+	frame_size = (uint32_t)tegra_ivc_total_queue_size(msg_size);
+	if (frame_size > TEGRA_BPMP_IPC_CH_MAP_SIZE) {
+		ERROR("%s: carveout size is not sufficient\n", __func__);
+		return -EINVAL;
+	}
+
+	error = tegra_ivc_init(&ivc_ccplex_bpmp_channel,
+				(uint32_t)TEGRA_BPMP_IPC_RX_PHYS_BASE,
+				(uint32_t)TEGRA_BPMP_IPC_TX_PHYS_BASE,
+				1U, frame_size, tegra_bpmp_ivc_notify);
+	if (error != 0) {
+
+		ERROR("%s: IVC init failed (%d)\n", __func__, error);
+
+	} else {
+
+		/* reset channel */
+		tegra_ivc_channel_reset(&ivc_ccplex_bpmp_channel);
+
+		/* wait for notification from BPMP */
+		while (tegra_ivc_channel_notified(&ivc_ccplex_bpmp_channel) != 0) {
+			/*
+			 * Interrupt BPMP with doorbell each time after
+			 * tegra_ivc_channel_notified() returns non zero
+			 * value.
+			 */
+			tegra_bpmp_ring_bpmp_doorbell();
+		}
+
+		INFO("%s: All communication channels initialized\n", __func__);
+	}
+
+	return error;
+}
+
+/* Handler to reset a hardware module */
+int32_t tegra_bpmp_ipc_reset_module(uint32_t rst_id)
+{
+	int32_t ret;
+	struct mrq_reset_request req = {
+		.cmd = (uint32_t)CMD_RESET_MODULE,
+		.reset_id = rst_id
+	};
+
+	/* only GPCDMA/XUSB_PADCTL resets are supported */
+	assert((rst_id == TEGRA_RESET_ID_XUSB_PADCTL) ||
+	       (rst_id == TEGRA_RESET_ID_GPCDMA));
+
+	ret = tegra_bpmp_ipc_send_req_atomic(MRQ_RESET, &req,
+			(uint32_t)sizeof(req), NULL, 0);
+	if (ret != 0) {
+		ERROR("%s: failed for module %d with error %d\n", __func__,
+		      rst_id, ret);
+	}
+
+	return ret;
+}
diff --git a/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.h b/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.h
new file mode 100644
index 0000000..689f8bb
--- /dev/null
+++ b/plat/nvidia/tegra/common/drivers/bpmp_ipc/intf.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#ifndef INTF_H
+#define INTF_H
+
+/**
+ * Flags used in IPC req
+ */
+#define FLAG_DO_ACK			(U(1) << 0)
+#define FLAG_RING_DOORBELL		(U(1) << 1)
+
+/* Bit 1 is designated for CCPlex in secure world */
+#define HSP_MASTER_CCPLEX_BIT		(U(1) << 1)
+/* Bit 19 is designated for BPMP in non-secure world */
+#define HSP_MASTER_BPMP_BIT		(U(1) << 19)
+/* Timeout to receive response from BPMP is 1 sec */
+#define TIMEOUT_RESPONSE_FROM_BPMP_US	U(1000000) /* in microseconds */
+
+/**
+ * IVC protocol defines and command/response frame
+ */
+
+/**
+ * IVC specific defines
+ */
+#define IVC_CMD_SZ_BYTES		U(128)
+#define IVC_DATA_SZ_BYTES		U(120)
+
+/**
+ * Holds frame data for an IPC request
+ */
+struct frame_data {
+	/* Identification as to what kind of data is being transmitted */
+	uint32_t mrq;
+
+	/* Flags for slave as to how to respond back */
+	uint32_t flags;
+
+	/* Actual data being sent */
+	uint8_t data[IVC_DATA_SZ_BYTES];
+};
+
+/**
+ * Commands send to the BPMP firmware
+ */
+
+/**
+ * MRQ code to issue a module reset command to BPMP
+ */
+#define MRQ_RESET			U(20)
+
+/**
+ * Reset sub-commands
+ */
+#define CMD_RESET_ASSERT		U(1)
+#define CMD_RESET_DEASSERT		U(2)
+#define CMD_RESET_MODULE		U(3)
+
+/**
+ * Used by the sender of an #MRQ_RESET message to request BPMP to
+ * assert or deassert a given reset line.
+ */
+struct __attribute__((packed)) mrq_reset_request {
+	/* reset action to perform (mrq_reset_commands) */
+	uint32_t cmd;
+	/* id of the reset to affected */
+	uint32_t reset_id;
+};
+
+#endif /* INTF_H */
diff --git a/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.c b/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.c
new file mode 100644
index 0000000..4212eca
--- /dev/null
+++ b/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.c
@@ -0,0 +1,653 @@
+/*
+ * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#include <arch_helpers.h>
+#include <assert.h>
+#include <debug.h>
+#include <errno.h>
+#include <stddef.h>
+#include <string.h>
+
+#include "ivc.h"
+
+/*
+ * IVC channel reset protocol.
+ *
+ * Each end uses its tx_channel.state to indicate its synchronization state.
+ */
+enum {
+	/*
+	 * This value is zero for backwards compatibility with services that
+	 * assume channels to be initially zeroed. Such channels are in an
+	 * initially valid state, but cannot be asynchronously reset, and must
+	 * maintain a valid state at all times.
+	 *
+	 * The transmitting end can enter the established state from the sync or
+	 * ack state when it observes the receiving endpoint in the ack or
+	 * established state, indicating that has cleared the counters in our
+	 * rx_channel.
+	 */
+	ivc_state_established = U(0),
+
+	/*
+	 * If an endpoint is observed in the sync state, the remote endpoint is
+	 * allowed to clear the counters it owns asynchronously with respect to
+	 * the current endpoint. Therefore, the current endpoint is no longer
+	 * allowed to communicate.
+	 */
+	ivc_state_sync = U(1),
+
+	/*
+	 * When the transmitting end observes the receiving end in the sync
+	 * state, it can clear the w_count and r_count and transition to the ack
+	 * state. If the remote endpoint observes us in the ack state, it can
+	 * return to the established state once it has cleared its counters.
+	 */
+	ivc_state_ack = U(2)
+};
+
+/*
+ * This structure is divided into two-cache aligned parts, the first is only
+ * written through the tx_channel pointer, while the second is only written
+ * through the rx_channel pointer. This delineates ownership of the cache lines,
+ * which is critical to performance and necessary in non-cache coherent
+ * implementations.
+ */
+struct ivc_channel_header {
+	struct {
+		/* fields owned by the transmitting end */
+		uint32_t w_count;
+		uint32_t state;
+		uint32_t w_rsvd[IVC_CHHDR_TX_FIELDS - 2];
+	};
+	struct {
+		/* fields owned by the receiving end */
+		uint32_t r_count;
+		uint32_t r_rsvd[IVC_CHHDR_RX_FIELDS - 1];
+	};
+};
+
+static inline bool ivc_channel_empty(const struct ivc *ivc,
+		volatile const struct ivc_channel_header *ch)
+{
+	/*
+	 * This function performs multiple checks on the same values with
+	 * security implications, so sample the counters' current values in
+	 * shared memory to ensure that these checks use the same values.
+	 */
+	uint32_t wr_count = ch->w_count;
+	uint32_t rd_count = ch->r_count;
+	bool ret = false;
+
+	(void)ivc;
+
+	/*
+	 * Perform an over-full check to prevent denial of service attacks where
+	 * a server could be easily fooled into believing that there's an
+	 * extremely large number of frames ready, since receivers are not
+	 * expected to check for full or over-full conditions.
+	 *
+	 * Although the channel isn't empty, this is an invalid case caused by
+	 * a potentially malicious peer, so returning empty is safer, because it
+	 * gives the impression that the channel has gone silent.
+	 */
+	if (((wr_count - rd_count) > ivc->nframes) || (wr_count == rd_count)) {
+		ret = true;
+	}
+
+	return ret;
+}
+
+static inline bool ivc_channel_full(const struct ivc *ivc,
+		volatile const struct ivc_channel_header *ch)
+{
+	uint32_t wr_count = ch->w_count;
+	uint32_t rd_count = ch->r_count;
+
+	(void)ivc;
+
+	/*
+	 * Invalid cases where the counters indicate that the queue is over
+	 * capacity also appear full.
+	 */
+	return ((wr_count - rd_count) >= ivc->nframes);
+}
+
+static inline uint32_t ivc_channel_avail_count(const struct ivc *ivc,
+		volatile const struct ivc_channel_header *ch)
+{
+	uint32_t wr_count = ch->w_count;
+	uint32_t rd_count = ch->r_count;
+
+	(void)ivc;
+
+	/*
+	 * This function isn't expected to be used in scenarios where an
+	 * over-full situation can lead to denial of service attacks. See the
+	 * comment in ivc_channel_empty() for an explanation about special
+	 * over-full considerations.
+	 */
+	return (wr_count - rd_count);
+}
+
+static inline void ivc_advance_tx(struct ivc *ivc)
+{
+	ivc->tx_channel->w_count++;
+
+	if (ivc->w_pos == (ivc->nframes - (uint32_t)1U)) {
+		ivc->w_pos = 0U;
+	} else {
+		ivc->w_pos++;
+	}
+}
+
+static inline void ivc_advance_rx(struct ivc *ivc)
+{
+	ivc->rx_channel->r_count++;
+
+	if (ivc->r_pos == (ivc->nframes - (uint32_t)1U)) {
+		ivc->r_pos = 0U;
+	} else {
+		ivc->r_pos++;
+	}
+}
+
+static inline int32_t ivc_check_read(const struct ivc *ivc)
+{
+	/*
+	 * tx_channel->state is set locally, so it is not synchronized with
+	 * state from the remote peer. The remote peer cannot reset its
+	 * transmit counters until we've acknowledged its synchronization
+	 * request, so no additional synchronization is required because an
+	 * asynchronous transition of rx_channel->state to ivc_state_ack is not
+	 * allowed.
+	 */
+	if (ivc->tx_channel->state != ivc_state_established) {
+		return -ECONNRESET;
+	}
+
+	/*
+	* Avoid unnecessary invalidations when performing repeated accesses to
+	* an IVC channel by checking the old queue pointers first.
+	* Synchronization is only necessary when these pointers indicate empty
+	* or full.
+	*/
+	if (!ivc_channel_empty(ivc, ivc->rx_channel)) {
+		return 0;
+	}
+
+	return ivc_channel_empty(ivc, ivc->rx_channel) ? -ENOMEM : 0;
+}
+
+static inline int32_t ivc_check_write(const struct ivc *ivc)
+{
+	if (ivc->tx_channel->state != ivc_state_established) {
+		return -ECONNRESET;
+	}
+
+	if (!ivc_channel_full(ivc, ivc->tx_channel)) {
+		return 0;
+	}
+
+	return ivc_channel_full(ivc, ivc->tx_channel) ? -ENOMEM : 0;
+}
+
+bool tegra_ivc_can_read(const struct ivc *ivc)
+{
+	return ivc_check_read(ivc) == 0;
+}
+
+bool tegra_ivc_can_write(const struct ivc *ivc)
+{
+	return ivc_check_write(ivc) == 0;
+}
+
+bool tegra_ivc_tx_empty(const struct ivc *ivc)
+{
+	return ivc_channel_empty(ivc, ivc->tx_channel);
+}
+
+static inline uintptr_t calc_frame_offset(uint32_t frame_index,
+	uint32_t frame_size, uint32_t frame_offset)
+{
+    return ((uintptr_t)frame_index * (uintptr_t)frame_size) +
+	    (uintptr_t)frame_offset;
+}
+
+static void *ivc_frame_pointer(const struct ivc *ivc,
+				volatile const struct ivc_channel_header *ch,
+				uint32_t frame)
+{
+	assert(frame < ivc->nframes);
+	return (void *)((uintptr_t)(&ch[1]) +
+		calc_frame_offset(frame, ivc->frame_size, 0));
+}
+
+int32_t tegra_ivc_read(struct ivc *ivc, void *buf, size_t max_read)
+{
+	const void *src;
+	int32_t result;
+
+	if (buf == NULL) {
+		return -EINVAL;
+	}
+
+	if (max_read > ivc->frame_size) {
+		return -E2BIG;
+	}
+
+	result = ivc_check_read(ivc);
+	if (result != 0) {
+		return result;
+	}
+
+	/*
+	 * Order observation of w_pos potentially indicating new data before
+	 * data read.
+	 */
+	dmbish();
+
+	src = ivc_frame_pointer(ivc, ivc->rx_channel, ivc->r_pos);
+
+	(void)memcpy(buf, src, max_read);
+
+	ivc_advance_rx(ivc);
+
+	/*
+	 * Ensure our write to r_pos occurs before our read from w_pos.
+	 */
+	dmbish();
+
+	/*
+	 * Notify only upon transition from full to non-full.
+	 * The available count can only asynchronously increase, so the
+	 * worst possible side-effect will be a spurious notification.
+	 */
+	if (ivc_channel_avail_count(ivc, ivc->rx_channel) == (ivc->nframes - (uint32_t)1U)) {
+		ivc->notify(ivc);
+	}
+
+	return (int32_t)max_read;
+}
+
+/* directly peek at the next frame rx'ed */
+void *tegra_ivc_read_get_next_frame(const struct ivc *ivc)
+{
+	if (ivc_check_read(ivc) != 0) {
+		return NULL;
+	}
+
+	/*
+	 * Order observation of w_pos potentially indicating new data before
+	 * data read.
+	 */
+	dmbld();
+
+	return ivc_frame_pointer(ivc, ivc->rx_channel, ivc->r_pos);
+}
+
+int32_t tegra_ivc_read_advance(struct ivc *ivc)
+{
+	/*
+	 * No read barriers or synchronization here: the caller is expected to
+	 * have already observed the channel non-empty. This check is just to
+	 * catch programming errors.
+	 */
+	int32_t result = ivc_check_read(ivc);
+	if (result != 0) {
+		return result;
+	}
+
+	ivc_advance_rx(ivc);
+
+	/*
+	 * Ensure our write to r_pos occurs before our read from w_pos.
+	 */
+	dmbish();
+
+	/*
+	 * Notify only upon transition from full to non-full.
+	 * The available count can only asynchronously increase, so the
+	 * worst possible side-effect will be a spurious notification.
+	 */
+	if (ivc_channel_avail_count(ivc, ivc->rx_channel) == (ivc->nframes - (uint32_t)1U)) {
+		ivc->notify(ivc);
+	}
+
+	return 0;
+}
+
+int32_t tegra_ivc_write(struct ivc *ivc, const void *buf, size_t size)
+{
+	void *p;
+	int32_t result;
+
+	if ((buf == NULL) || (ivc == NULL)) {
+		return -EINVAL;
+	}
+
+	if (size > ivc->frame_size) {
+		return -E2BIG;
+	}
+
+	result = ivc_check_write(ivc);
+	if (result != 0) {
+		return result;
+	}
+
+	p = ivc_frame_pointer(ivc, ivc->tx_channel, ivc->w_pos);
+
+	(void)memset(p, 0, ivc->frame_size);
+	(void)memcpy(p, buf, size);
+
+	/*
+	 * Ensure that updated data is visible before the w_pos counter
+	 * indicates that it is ready.
+	 */
+	dmbst();
+
+	ivc_advance_tx(ivc);
+
+	/*
+	 * Ensure our write to w_pos occurs before our read from r_pos.
+	 */
+	dmbish();
+
+	/*
+	 * Notify only upon transition from empty to non-empty.
+	 * The available count can only asynchronously decrease, so the
+	 * worst possible side-effect will be a spurious notification.
+	 */
+	if (ivc_channel_avail_count(ivc, ivc->tx_channel) == 1U) {
+		ivc->notify(ivc);
+	}
+
+	return (int32_t)size;
+}
+
+/* directly poke at the next frame to be tx'ed */
+void *tegra_ivc_write_get_next_frame(const struct ivc *ivc)
+{
+	if (ivc_check_write(ivc) != 0) {
+		return NULL;
+	}
+
+	return ivc_frame_pointer(ivc, ivc->tx_channel, ivc->w_pos);
+}
+
+/* advance the tx buffer */
+int32_t tegra_ivc_write_advance(struct ivc *ivc)
+{
+	int32_t result = ivc_check_write(ivc);
+
+	if (result != 0) {
+		return result;
+	}
+
+	/*
+	 * Order any possible stores to the frame before update of w_pos.
+	 */
+	dmbst();
+
+	ivc_advance_tx(ivc);
+
+	/*
+	 * Ensure our write to w_pos occurs before our read from r_pos.
+	 */
+	dmbish();
+
+	/*
+	 * Notify only upon transition from empty to non-empty.
+	 * The available count can only asynchronously decrease, so the
+	 * worst possible side-effect will be a spurious notification.
+	 */
+	if (ivc_channel_avail_count(ivc, ivc->tx_channel) == (uint32_t)1U) {
+		ivc->notify(ivc);
+	}
+
+	return 0;
+}
+
+void tegra_ivc_channel_reset(const struct ivc *ivc)
+{
+	ivc->tx_channel->state = ivc_state_sync;
+	ivc->notify(ivc);
+}
+
+/*
+ * ===============================================================
+ *  IVC State Transition Table - see tegra_ivc_channel_notified()
+ * ===============================================================
+ *
+ *	local	remote	action
+ *	-----	------	-----------------------------------
+ *	SYNC	EST	<none>
+ *	SYNC	ACK	reset counters; move to EST; notify
+ *	SYNC	SYNC	reset counters; move to ACK; notify
+ *	ACK	EST	move to EST; notify
+ *	ACK	ACK	move to EST; notify
+ *	ACK	SYNC	reset counters; move to ACK; notify
+ *	EST	EST	<none>
+ *	EST	ACK	<none>
+ *	EST	SYNC	reset counters; move to ACK; notify
+ *
+ * ===============================================================
+ */
+int32_t tegra_ivc_channel_notified(struct ivc *ivc)
+{
+	uint32_t peer_state;
+
+	/* Copy the receiver's state out of shared memory. */
+	peer_state = ivc->rx_channel->state;
+
+	if (peer_state == (uint32_t)ivc_state_sync) {
+		/*
+		 * Order observation of ivc_state_sync before stores clearing
+		 * tx_channel.
+		 */
+		dmbld();
+
+		/*
+		 * Reset tx_channel counters. The remote end is in the SYNC
+		 * state and won't make progress until we change our state,
+		 * so the counters are not in use at this time.
+		 */
+		ivc->tx_channel->w_count = 0U;
+		ivc->rx_channel->r_count = 0U;
+
+		ivc->w_pos = 0U;
+		ivc->r_pos = 0U;
+
+		/*
+		 * Ensure that counters appear cleared before new state can be
+		 * observed.
+		 */
+		dmbst();
+
+		/*
+		 * Move to ACK state. We have just cleared our counters, so it
+		 * is now safe for the remote end to start using these values.
+		 */
+		ivc->tx_channel->state = ivc_state_ack;
+
+		/*
+		 * Notify remote end to observe state transition.
+		 */
+		ivc->notify(ivc);
+
+	} else if ((ivc->tx_channel->state == (uint32_t)ivc_state_sync) &&
+			(peer_state == (uint32_t)ivc_state_ack)) {
+		/*
+		 * Order observation of ivc_state_sync before stores clearing
+		 * tx_channel.
+		 */
+		dmbld();
+
+		/*
+		 * Reset tx_channel counters. The remote end is in the ACK
+		 * state and won't make progress until we change our state,
+		 * so the counters are not in use at this time.
+		 */
+		ivc->tx_channel->w_count = 0U;
+		ivc->rx_channel->r_count = 0U;
+
+		ivc->w_pos = 0U;
+		ivc->r_pos = 0U;
+
+		/*
+		 * Ensure that counters appear cleared before new state can be
+		 * observed.
+		 */
+		dmbst();
+
+		/*
+		 * Move to ESTABLISHED state. We know that the remote end has
+		 * already cleared its counters, so it is safe to start
+		 * writing/reading on this channel.
+		 */
+		ivc->tx_channel->state = ivc_state_established;
+
+		/*
+		 * Notify remote end to observe state transition.
+		 */
+		ivc->notify(ivc);
+
+	} else if (ivc->tx_channel->state == (uint32_t)ivc_state_ack) {
+		/*
+		 * At this point, we have observed the peer to be in either
+		 * the ACK or ESTABLISHED state. Next, order observation of
+		 * peer state before storing to tx_channel.
+		 */
+		dmbld();
+
+		/*
+		 * Move to ESTABLISHED state. We know that we have previously
+		 * cleared our counters, and we know that the remote end has
+		 * cleared its counters, so it is safe to start writing/reading
+		 * on this channel.
+		 */
+		ivc->tx_channel->state = ivc_state_established;
+
+		/*
+		 * Notify remote end to observe state transition.
+		 */
+		ivc->notify(ivc);
+
+	} else {
+		/*
+		 * There is no need to handle any further action. Either the
+		 * channel is already fully established, or we are waiting for
+		 * the remote end to catch up with our current state. Refer
+		 * to the diagram in "IVC State Transition Table" above.
+		 */
+	}
+
+	return ((ivc->tx_channel->state == (uint32_t)ivc_state_established) ? 0 : -EAGAIN);
+}
+
+size_t tegra_ivc_align(size_t size)
+{
+	return (size + (IVC_ALIGN - 1U)) & ~(IVC_ALIGN - 1U);
+}
+
+size_t tegra_ivc_total_queue_size(size_t queue_size)
+{
+	if ((queue_size & (IVC_ALIGN - 1U)) != 0U) {
+		ERROR("queue_size (%d) must be %d-byte aligned\n",
+				(int32_t)queue_size, IVC_ALIGN);
+		return 0;
+	}
+	return queue_size + sizeof(struct ivc_channel_header);
+}
+
+static int32_t check_ivc_params(uintptr_t queue_base1, uintptr_t queue_base2,
+		uint32_t nframes, uint32_t frame_size)
+{
+	assert((offsetof(struct ivc_channel_header, w_count)
+				& (IVC_ALIGN - 1U)) == 0U);
+	assert((offsetof(struct ivc_channel_header, r_count)
+				& (IVC_ALIGN - 1U)) == 0U);
+	assert((sizeof(struct ivc_channel_header) & (IVC_ALIGN - 1U)) == 0U);
+
+	if (((uint64_t)nframes * (uint64_t)frame_size) >= 0x100000000ULL) {
+		ERROR("nframes * frame_size overflows\n");
+		return -EINVAL;
+	}
+
+	/*
+	 * The headers must at least be aligned enough for counters
+	 * to be accessed atomically.
+	 */
+	if ((queue_base1 & (IVC_ALIGN - 1U)) != 0U) {
+		ERROR("ivc channel start not aligned: %lx\n", queue_base1);
+		return -EINVAL;
+	}
+	if ((queue_base2 & (IVC_ALIGN - 1U)) != 0U) {
+		ERROR("ivc channel start not aligned: %lx\n", queue_base2);
+		return -EINVAL;
+	}
+
+	if ((frame_size & (IVC_ALIGN - 1U)) != 0U) {
+		ERROR("frame size not adequately aligned: %u\n",
+				frame_size);
+		return -EINVAL;
+	}
+
+	if (queue_base1 < queue_base2) {
+		if ((queue_base1 + ((uint64_t)frame_size * nframes)) > queue_base2) {
+			ERROR("queue regions overlap: %lx + %x, %x\n",
+					queue_base1, frame_size,
+					frame_size * nframes);
+			return -EINVAL;
+		}
+	} else {
+		if ((queue_base2 + ((uint64_t)frame_size * nframes)) > queue_base1) {
+			ERROR("queue regions overlap: %lx + %x, %x\n",
+					queue_base2, frame_size,
+					frame_size * nframes);
+			return -EINVAL;
+		}
+	}
+
+	return 0;
+}
+
+int32_t tegra_ivc_init(struct ivc *ivc, uintptr_t rx_base, uintptr_t tx_base,
+		uint32_t nframes, uint32_t frame_size,
+		ivc_notify_function notify)
+{
+	int32_t result;
+
+	/* sanity check input params */
+	if ((ivc == NULL) || (notify == NULL)) {
+		return -EINVAL;
+	}
+
+	result = check_ivc_params(rx_base, tx_base, nframes, frame_size);
+	if (result != 0) {
+		return result;
+	}
+
+	/*
+	 * All sizes that can be returned by communication functions should
+	 * fit in a 32-bit integer.
+	 */
+	if (frame_size > (1u << 31)) {
+		return -E2BIG;
+	}
+
+	ivc->rx_channel = (struct ivc_channel_header *)rx_base;
+	ivc->tx_channel = (struct ivc_channel_header *)tx_base;
+	ivc->notify = notify;
+	ivc->frame_size = frame_size;
+	ivc->nframes = nframes;
+	ivc->w_pos = 0U;
+	ivc->r_pos = 0U;
+
+	INFO("%s: done\n", __func__);
+
+	return 0;
+}
diff --git a/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.h b/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.h
new file mode 100644
index 0000000..f34d6cf
--- /dev/null
+++ b/plat/nvidia/tegra/common/drivers/bpmp_ipc/ivc.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#ifndef IVC_H
+#define IVC_H
+
+#include <stdint.h>
+#include <stddef.h>
+#include <utils_def.h>
+
+#define IVC_ALIGN		U(64)
+#define IVC_CHHDR_TX_FIELDS	U(16)
+#define IVC_CHHDR_RX_FIELDS	U(16)
+
+struct ivc;
+struct ivc_channel_header;
+
+/* callback handler for notify on receiving a response */
+typedef void (* ivc_notify_function)(const struct ivc *);
+
+struct ivc {
+	struct ivc_channel_header *rx_channel;
+	struct ivc_channel_header *tx_channel;
+	uint32_t w_pos;
+	uint32_t r_pos;
+	ivc_notify_function notify;
+	uint32_t nframes;
+	uint32_t frame_size;
+};
+
+int32_t tegra_ivc_init(struct ivc *ivc, uintptr_t rx_base, uintptr_t tx_base,
+		uint32_t nframes, uint32_t frame_size,
+		ivc_notify_function notify);
+size_t tegra_ivc_total_queue_size(size_t queue_size);
+size_t tegra_ivc_align(size_t size);
+int32_t tegra_ivc_channel_notified(struct ivc *ivc);
+void tegra_ivc_channel_reset(const struct ivc *ivc);
+int32_t tegra_ivc_write_advance(struct ivc *ivc);
+void *tegra_ivc_write_get_next_frame(const struct ivc *ivc);
+int32_t tegra_ivc_write(struct ivc *ivc, const void *buf, size_t size);
+int32_t tegra_ivc_read_advance(struct ivc *ivc);
+void *tegra_ivc_read_get_next_frame(const struct ivc *ivc);
+int32_t tegra_ivc_read(struct ivc *ivc, void *buf, size_t max_read);
+bool tegra_ivc_tx_empty(const struct ivc *ivc);
+bool tegra_ivc_can_write(const struct ivc *ivc);
+bool tegra_ivc_can_read(const struct ivc *ivc);
+
+#endif /* IVC_H */
diff --git a/plat/nvidia/tegra/include/drivers/bpmp_ipc.h b/plat/nvidia/tegra/include/drivers/bpmp_ipc.h
new file mode 100644
index 0000000..9304150
--- /dev/null
+++ b/plat/nvidia/tegra/include/drivers/bpmp_ipc.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ *
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+
+#ifndef __BPMP_IPC_H__
+#define __BPMP_IPC_H__
+
+#include <stdbool.h>
+#include <stdint.h>
+#include <utils_def.h>
+
+/**
+ * Currently supported reset identifiers
+ */
+#define TEGRA_RESET_ID_XUSB_PADCTL	U(114)
+#define TEGRA_RESET_ID_GPCDMA		U(70)
+
+/**
+ * Function to initialise the IPC with the bpmp
+ */
+int32_t tegra_bpmp_ipc_init(void);
+
+/**
+ * Handler to reset a module
+ */
+int32_t tegra_bpmp_ipc_reset_module(uint32_t rst_id);
+
+#endif /* __BPMP_IPC_H__ */