Re: [PATCH v3 05/12] firmware: tegra: Add BPMP support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Aug 19, 2016 at 07:32:26PM +0200, Thierry Reding wrote:
> From: Thierry Reding <treding@xxxxxxxxxx>
> 
> The Boot and Power Management Processor (BPMP) is a co-processor found
> on Tegra SoCs. It is designed to handle the early stages of the boot
> process and offload power management tasks (such as clocks, resets,
> powergates, ...) as well as system control services.
> 
> Compared to the ARM SCPI, the services provided by BPMP are message-
> based rather than method-based. The BPMP firmware driver provides the
> services to transmit data to and receive data from the BPMP. Users can
> also register an MRQ, for which a service routine will be run when a
> corresponding event is received from the firmware.
> 
> A set of messages, called the BPMP ABI, are specified for a number of
> different services provided by the BPMP (such as clocks or resets).
> 
> Based on work by Sivaram Nair <sivaramn@xxxxxxxxxx> and Joseph Lo
> <josephl@xxxxxxxxxx>.
> 
> Signed-off-by: Thierry Reding <treding@xxxxxxxxxx>
> ---
> Changes in v3:
> - use a message structure for transfers to avoid long function
>   prototypes
> - restructure driver for easier maintainability
> - rename bpmp_abi.h to bpmp-abi.h for consistency
> 
>  drivers/firmware/tegra/Kconfig  |   12 +
>  drivers/firmware/tegra/Makefile |    1 +
>  drivers/firmware/tegra/bpmp.c   |  880 +++++++++++++++++++++
>  include/soc/tegra/bpmp-abi.h    | 1601 +++++++++++++++++++++++++++++++++++++++
>  include/soc/tegra/bpmp.h        |  122 +++
>  5 files changed, 2616 insertions(+)
>  create mode 100644 drivers/firmware/tegra/bpmp.c
>  create mode 100644 include/soc/tegra/bpmp-abi.h
>  create mode 100644 include/soc/tegra/bpmp.h
> 
> diff --git a/drivers/firmware/tegra/Kconfig b/drivers/firmware/tegra/Kconfig
> index 1fa3e4e136a5..ff2730d5c468 100644
> --- a/drivers/firmware/tegra/Kconfig
> +++ b/drivers/firmware/tegra/Kconfig
> @@ -10,4 +10,16 @@ config TEGRA_IVC
>  	  keeps the content is synchronization between host CPU and remote
>  	  processors.
>  
> +config TEGRA_BPMP
> +	bool "Tegra BPMP driver"
> +	depends on ARCH_TEGRA && TEGRA_HSP_MBOX && TEGRA_IVC
> +	help
> +	  BPMP (Boot and Power Management Processor) is designed to off-loading
> +	  the PM functions which include clock/DVFS/thermal/power from the CPU.
> +	  It needs HSP as the HW synchronization and notification module and
> +	  IVC module as the message communication protocol.
> +
> +	  This driver manages the IPC interface between host CPU and the
> +	  firmware running on BPMP.
> +
>  endmenu
> diff --git a/drivers/firmware/tegra/Makefile b/drivers/firmware/tegra/Makefile
> index 92e2153e8173..e34a2f79e1ad 100644
> --- a/drivers/firmware/tegra/Makefile
> +++ b/drivers/firmware/tegra/Makefile
> @@ -1 +1,2 @@
> +obj-$(CONFIG_TEGRA_BPMP)	+= bpmp.o
>  obj-$(CONFIG_TEGRA_IVC)		+= ivc.o
> diff --git a/drivers/firmware/tegra/bpmp.c b/drivers/firmware/tegra/bpmp.c
> new file mode 100644
> index 000000000000..a09043b1be05
> --- /dev/null
> +++ b/drivers/firmware/tegra/bpmp.c
> @@ -0,0 +1,880 @@
> +/*
> + * Copyright (c) 2016, NVIDIA CORPORATION.  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.
> + */
> +
> +#define DEBUG
> +
> +#include <linux/clk/tegra.h>
> +#include <linux/mailbox_client.h>
> +#include <linux/of.h>
> +#include <linux/of_address.h>
> +#include <linux/of_device.h>
> +#include <linux/platform_device.h>
> +#include <linux/semaphore.h>
> +
> +#include <soc/tegra/bpmp.h>
> +#include <soc/tegra/bpmp-abi.h>
> +#include <soc/tegra/ivc.h>
> +
> +#define MSG_ACK		BIT(0)
> +#define MSG_RING	BIT(1)
> +
> +static inline struct tegra_bpmp *
> +mbox_client_to_bpmp(struct mbox_client *client)
> +{
> +	return container_of(client, struct tegra_bpmp, mbox.client);
> +}
> +
> +struct tegra_bpmp *tegra_bpmp_get(struct device *dev)
> +{
> +	struct platform_device *pdev;
> +	struct tegra_bpmp *bpmp;
> +	struct device_node *np;
> +
> +	np = of_parse_phandle(dev->of_node, "nvidia,bpmp", 0);
> +	if (!np)
> +		return ERR_PTR(-ENOENT);
> +
> +	pdev = of_find_device_by_node(np);
> +	if (!pdev) {
> +		bpmp = ERR_PTR(-ENODEV);
> +		goto put;
> +	}
> +
> +	bpmp = platform_get_drvdata(pdev);
> +	if (!bpmp) {
> +		bpmp = ERR_PTR(-EPROBE_DEFER);
> +		put_device(&pdev->dev);
> +		goto put;
> +	}
> +
> +put:
> +	of_node_put(np);
> +	return bpmp;
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_get);
> +
> +void tegra_bpmp_put(struct tegra_bpmp *bpmp)
> +{
> +	if (bpmp)
> +		put_device(bpmp->dev);
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_put);
> +
> +static int tegra_bpmp_channel_get_index(struct tegra_bpmp_channel *channel)
> +{
> +	return channel - channel->bpmp->channels;
> +}
> +
> +static int
> +tegra_bpmp_channel_get_thread_index(struct tegra_bpmp_channel *channel)
> +{
> +	struct tegra_bpmp *bpmp = channel->bpmp;
> +	unsigned int offset, count;
> +	int index;
> +
> +	offset = bpmp->soc->channels.thread.offset;
> +	count = bpmp->soc->channels.thread.count;
> +
> +	index = tegra_bpmp_channel_get_index(channel);
> +	if (index < 0)
> +		return index;
> +
> +	if (index < offset || index >= offset + count)
> +		return -EINVAL;
> +
> +	return index - offset;
> +}
> +
> +static struct tegra_bpmp_channel *
> +tegra_bpmp_channel_get_thread(struct tegra_bpmp *bpmp, unsigned int index)
> +{
> +	unsigned int offset = bpmp->soc->channels.thread.offset;
> +	unsigned int count = bpmp->soc->channels.thread.count;
> +
> +	if (index >= count)
> +		return NULL;
> +
> +	return &bpmp->channels[offset + index];
> +}
> +
> +static struct tegra_bpmp_channel *
> +tegra_bpmp_channel_get_tx(struct tegra_bpmp *bpmp)
> +{
> +	unsigned int offset = bpmp->soc->channels.cpu_tx.offset;
> +
> +	return &bpmp->channels[offset + smp_processor_id()];
> +}
> +
> +static struct tegra_bpmp_channel *
> +tegra_bpmp_channel_get_rx(struct tegra_bpmp *bpmp)
> +{
> +	unsigned int offset = bpmp->soc->channels.cpu_rx.offset;
> +
> +	return &bpmp->channels[offset];
> +}
> +
> +static bool tegra_bpmp_message_valid(const struct tegra_bpmp_message *msg)
> +{
> +	return (msg->tx.size <= TEGRA_BPMP_MSG_DATA_SIZE) &&
> +	       (msg->rx.size <= TEGRA_BPMP_MSG_DATA_SIZE) &&
> +	       (msg->tx.size == 0 || msg->tx.data) &&
> +	       (msg->rx.size == 0 || msg->rx.data);
> +}
> +
> +static bool tegra_bpmp_master_acked(struct tegra_bpmp_channel *channel)
> +{
> +	void *frame;
> +
> +	frame = tegra_ivc_read_get_next_frame(channel->ivc);
> +	if (IS_ERR_OR_NULL(frame)) {
> +		channel->ib = NULL;
> +		return false;
> +	}
> +
> +	channel->ib = frame;
> +
> +	return true;
> +}
> +
> +static int tegra_bpmp_wait_ack(struct tegra_bpmp_channel *channel)
> +{
> +	unsigned long timeout = channel->bpmp->soc->channels.cpu_tx.timeout;
> +	ktime_t start, now;
> +
> +	start = ns_to_ktime(local_clock());
> +
> +	do {
> +		if (tegra_bpmp_master_acked(channel))
> +			return 0;
> +
> +		now = ns_to_ktime(local_clock());
> +	} while (ktime_us_delta(now, start) < timeout);
> +
> +	return -ETIMEDOUT;
> +}
> +
> +static bool tegra_bpmp_master_free(struct tegra_bpmp_channel *channel)
> +{
> +	void *frame;
> +
> +	frame = tegra_ivc_write_get_next_frame(channel->ivc);
> +	if (IS_ERR_OR_NULL(frame)) {
> +		channel->ob = NULL;
> +		return false;
> +	}
> +
> +	channel->ob = frame;
> +
> +	return true;
> +}
> +
> +static int tegra_bpmp_wait_master_free(struct tegra_bpmp_channel *channel)
> +{
> +	unsigned long timeout = channel->bpmp->soc->channels.cpu_tx.timeout;
> +	ktime_t start, now;
> +
> +	start = ns_to_ktime(local_clock());
> +
> +	do {
> +		if (tegra_bpmp_master_free(channel))
> +			return 0;
> +
> +		now = ns_to_ktime(local_clock());
> +	} while (ktime_us_delta(now, start) < timeout);
> +
> +	return -ETIMEDOUT;
> +}
> +
> +static ssize_t __tegra_bpmp_channel_read(struct tegra_bpmp_channel *channel,
> +					 void *data, size_t size)
> +{
> +	if (data && size > 0)
> +		memcpy_fromio(data, channel->ib->data, size);
> +
> +	return tegra_ivc_read_advance(channel->ivc);
> +}
> +
> +static ssize_t tegra_bpmp_channel_read(struct tegra_bpmp_channel *channel,
> +				       void *data, size_t size)
> +{
> +	struct tegra_bpmp *bpmp = channel->bpmp;
> +	unsigned long flags;
> +	ssize_t err;
> +	int index;
> +
> +	index = tegra_bpmp_channel_get_thread_index(channel);
> +	if (index < 0)
> +		return index;
> +
> +	spin_lock_irqsave(&bpmp->lock, flags);
> +	err = __tegra_bpmp_channel_read(channel, data, size);
> +	clear_bit(index, bpmp->threaded.allocated);
> +	spin_unlock_irqrestore(&bpmp->lock, flags);
> +
> +	up(&bpmp->threaded.lock);
> +
> +	return err;
> +}
> +
> +static ssize_t __tegra_bpmp_channel_write(struct tegra_bpmp_channel *channel,
> +					  unsigned int mrq, unsigned long flags,
> +					  const void *data, size_t size)
> +{
> +	channel->ob->code = mrq;
> +	channel->ob->flags = flags;
> +
> +	if (data && size > 0)
> +		memcpy_toio(channel->ob->data, data, size);
> +
> +	return tegra_ivc_write_advance(channel->ivc);
> +}
> +
> +static struct tegra_bpmp_channel *
> +tegra_bpmp_write_threaded(struct tegra_bpmp *bpmp, unsigned int mrq,
> +			  const void *data, size_t size)
> +{
> +	unsigned long timeout = bpmp->soc->channels.thread.timeout;
> +	unsigned int count = bpmp->soc->channels.thread.count;
> +	struct tegra_bpmp_channel *channel;
> +	unsigned long flags;
> +	unsigned int index;
> +	int err;
> +
> +	err = down_timeout(&bpmp->threaded.lock, usecs_to_jiffies(timeout));
> +	if (err < 0)
> +		return ERR_PTR(err);
> +
> +	spin_lock_irqsave(&bpmp->lock, flags);
> +
> +	index = find_first_zero_bit(bpmp->threaded.allocated, count);
> +	if (index == count) {
> +		channel = ERR_PTR(-EBUSY);
> +		goto unlock;
> +	}
> +
> +	channel = tegra_bpmp_channel_get_thread(bpmp, index);
> +	if (!channel) {
> +		channel = ERR_PTR(-EINVAL);
> +		goto unlock;
> +	}
> +
> +	if (!tegra_bpmp_master_free(channel)) {
> +		channel = ERR_PTR(-EBUSY);
> +		goto unlock;
> +	}
> +
> +	set_bit(index, bpmp->threaded.allocated);
> +
> +	err = __tegra_bpmp_channel_write(channel, mrq, MSG_ACK | MSG_RING,
> +					 data, size);
> +	if (err < 0) {
> +		clear_bit(index, bpmp->threaded.allocated);
> +		goto unlock;
> +	}
> +
> +	set_bit(index, bpmp->threaded.busy);
> +
> +unlock:
> +	spin_unlock_irqrestore(&bpmp->lock, flags);
> +	return channel;
> +}
> +
> +static ssize_t tegra_bpmp_channel_write(struct tegra_bpmp_channel *channel,
> +					unsigned int mrq, unsigned long flags,
> +					const void *data, size_t size)
> +{
> +	int err;
> +
> +	err = tegra_bpmp_wait_master_free(channel);
> +	if (err < 0)
> +		return err;
> +
> +	return __tegra_bpmp_channel_write(channel, mrq, flags, data, size);
> +}
> +
> +int tegra_bpmp_transfer_atomic(struct tegra_bpmp *bpmp,
> +			       struct tegra_bpmp_message *msg)
> +{
> +	struct tegra_bpmp_channel *channel;
> +	int err;
> +
> +	if (WARN_ON(!irqs_disabled()))
> +		return -EPERM;
> +
> +	if (!tegra_bpmp_message_valid(msg))
> +		return -EINVAL;
> +
> +	channel = tegra_bpmp_channel_get_tx(bpmp);
> +
> +	err = tegra_bpmp_channel_write(channel, msg->mrq, MSG_ACK,
> +				       msg->tx.data, msg->tx.size);
> +	if (err < 0)
> +		return err;
> +
> +	err = mbox_send_message(bpmp->mbox.channel, NULL);
> +	if (err < 0)
> +		return err;
> +
> +	mbox_client_txdone(bpmp->mbox.channel, 0);
> +
> +	err = tegra_bpmp_wait_ack(channel);
> +	if (err)
> +		return err;
> +
> +	return __tegra_bpmp_channel_read(channel, msg->rx.data, msg->rx.size);
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_transfer_atomic);
> +
> +int tegra_bpmp_transfer(struct tegra_bpmp *bpmp,
> +			struct tegra_bpmp_message *msg)
> +{
> +	struct tegra_bpmp_channel *channel;
> +	unsigned long timeout;
> +	int err;
> +
> +	if (WARN_ON(irqs_disabled()))
> +		return -EPERM;
> +
> +	if (!tegra_bpmp_message_valid(msg))
> +		return -EINVAL;
> +
> +	channel = tegra_bpmp_write_threaded(bpmp, msg->mrq, msg->tx.data,
> +					    msg->tx.size);
> +	if (IS_ERR(channel))
> +		return PTR_ERR(channel);
> +
> +	err = mbox_send_message(bpmp->mbox.channel, NULL);
> +	if (err < 0)
> +		return err;
> +
> +	mbox_client_txdone(bpmp->mbox.channel, 0);
> +
> +	timeout = usecs_to_jiffies(bpmp->soc->channels.thread.timeout);
> +
> +	err = wait_for_completion_timeout(&channel->completion, timeout);
> +	if (err <= 0) {

wait_for_completion_timeout() returns unsigned long - why are we
checking for < 0? The documentation says that zero indicates timeout.

> +		if (err == 0)
> +			err = -ETIMEDOUT;
> +
> +		return err;
> +	}
> +
> +	return tegra_bpmp_channel_read(channel, msg->rx.data, msg->rx.size);
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_transfer);
> +
> +static struct tegra_bpmp_mrq *tegra_bpmp_find_mrq(struct tegra_bpmp *bpmp,
> +						  unsigned int mrq)
> +{
> +	struct tegra_bpmp_mrq *entry;
> +
> +	list_for_each_entry(entry, &bpmp->mrqs, list)
> +		if (entry->mrq == mrq)
> +			return entry;
> +
> +	return NULL;
> +}
> +
> +static void tegra_bpmp_mrq_return(struct tegra_bpmp_channel *channel,
> +				  int code, const void *data, size_t size)
> +{
> +	unsigned long flags = channel->ib->flags;
> +	struct tegra_bpmp *bpmp = channel->bpmp;
> +	struct tegra_bpmp_mb_data *frame;
> +	int err;
> +
> +	if (WARN_ON(size > TEGRA_BPMP_MSG_DATA_SIZE))
> +		return;
> +
> +	err = tegra_ivc_read_advance(channel->ivc);
> +	if (WARN_ON(err < 0))
> +		return;
> +
> +	if ((flags & MSG_ACK) == 0)
> +		return;
> +
> +	frame = tegra_ivc_write_get_next_frame(channel->ivc);
> +	if (WARN_ON(IS_ERR_OR_NULL(frame)))
> +		return;
> +
> +	frame->code = code;
> +
> +	if (data && size > 0)
> +		memcpy_toio(frame->data, data, size);
> +
> +	err = tegra_ivc_write_advance(channel->ivc);
> +	if (WARN_ON(err < 0))
> +		return;
> +
> +	if (flags & MSG_RING) {
> +		err = mbox_send_message(bpmp->mbox.channel, NULL);
> +		if (WARN_ON(err < 0))
> +			return;
> +
> +		mbox_client_txdone(bpmp->mbox.channel, 0);
> +	}
> +}
> +
> +static void tegra_bpmp_handle_mrq(struct tegra_bpmp *bpmp,
> +				  unsigned int mrq,
> +				  struct tegra_bpmp_channel *channel)
> +{
> +	struct tegra_bpmp_mrq *entry;
> +	u32 zero = 0;
> +
> +	spin_lock(&bpmp->lock);
> +
> +	entry = tegra_bpmp_find_mrq(bpmp, mrq);
> +	if (!entry) {
> +		spin_unlock(&bpmp->lock);
> +		tegra_bpmp_mrq_return(channel, -EINVAL, &zero, sizeof(zero));
> +		return;
> +	}
> +
> +	entry->handler(mrq, channel, entry->data);
> +
> +	spin_unlock(&bpmp->lock);
> +}
> +
> +int tegra_bpmp_request_mrq(struct tegra_bpmp *bpmp, unsigned int mrq,
> +			   tegra_bpmp_mrq_handler_t handler, void *data)
> +{
> +	struct tegra_bpmp_mrq *entry;
> +	unsigned long flags;
> +
> +	if (!handler)
> +		return -EINVAL;
> +
> +	entry = devm_kzalloc(bpmp->dev, sizeof(*entry), GFP_KERNEL);
> +	if (!entry)
> +		return -ENOMEM;
> +
> +	spin_lock_irqsave(&bpmp->lock, flags);
> +
> +	entry->mrq = mrq;
> +	entry->handler = handler;
> +	entry->data = data;
> +	list_add(&entry->list, &bpmp->mrqs);
> +
> +	spin_unlock_irqrestore(&bpmp->lock, flags);
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_request_mrq);
> +
> +void tegra_bpmp_free_mrq(struct tegra_bpmp *bpmp, unsigned int mrq, void *data)
> +{
> +	struct tegra_bpmp_mrq *entry;
> +	unsigned long flags;
> +
> +	spin_lock_irqsave(&bpmp->lock, flags);
> +
> +	entry = tegra_bpmp_find_mrq(bpmp, mrq);
> +	if (!entry)
> +		goto unlock;
> +
> +	list_del(&entry->list);
> +	devm_kfree(bpmp->dev, entry);
> +
> +unlock:
> +	spin_unlock_irqrestore(&bpmp->lock, flags);
> +}
> +EXPORT_SYMBOL_GPL(tegra_bpmp_free_mrq);
> +
> +static void tegra_bpmp_mrq_handle_ping(unsigned int mrq,
> +				       struct tegra_bpmp_channel *channel,
> +				       void *data)
> +{
> +	struct mrq_ping_request *request;
> +	struct mrq_ping_response response;
> +
> +	request = (struct mrq_ping_request *)channel->ib->data;
> +
> +	memset(&response, 0, sizeof(response));

Why memset() - here and other places - ? Are you preparing for a
future/unforseen change to the ABI structures? If yes, I think we can ignore
that case - the existing structures should never change.

> +	response.reply = request->challenge << (smp_processor_id() + 1);

This does not follow the spec. reply should be challenge << 1
(downstream implementation seems wrong).

> +
> +	tegra_bpmp_mrq_return(channel, 0, &response, sizeof(response));
> +}
> +
> +static int tegra_bpmp_ping(struct tegra_bpmp *bpmp)
> +{
> +	struct mrq_ping_response response;
> +	struct mrq_ping_request request;
> +	struct tegra_bpmp_message msg;
> +	ktime_t start, delta;
> +	unsigned long flags;
> +	int err;
> +
> +	memset(&request, 0, sizeof(request));
> +	request.challenge = 1;
> +
> +	memset(&response, 0, sizeof(response));
> +
> +	memset(&msg, 0, sizeof(msg));
> +	msg.mrq = MRQ_PING;
> +	msg.tx.data = &request;
> +	msg.tx.size = sizeof(request);
> +	msg.rx.data = &response;
> +	msg.rx.size = sizeof(response);
> +
> +	start = ktime_get();

The ktime_get()s should be within the local_irq_save/restore() - so we
get a better measure of the round trip delay?

> +
> +	local_irq_save(flags);
> +	err = tegra_bpmp_transfer_atomic(bpmp, &msg);
> +	local_irq_restore(flags);
> +
> +	delta = ktime_sub(ktime_get(), start);
> +
> +	if (!err)
> +		dev_info(bpmp->dev,
> +			 "ping ok: challenge: %u, response: %u, time: %lld\n",
> +			 request.challenge, response.reply,
> +			 ktime_to_us(delta));
> +
> +	return err;
> +}
> +
> +static int tegra_bpmp_get_firmware_tag(struct tegra_bpmp *bpmp, char *tag,
> +				       size_t size)
> +{
> +	struct mrq_query_tag_request request;
> +	struct tegra_bpmp_message msg;
> +	unsigned long flags;
> +	dma_addr_t phys;
> +	void *virt;
> +	int err;
> +
> +	virt = dma_alloc_coherent(bpmp->dev, TEGRA_BPMP_MSG_DATA_SIZE, &phys,
> +				  GFP_KERNEL | GFP_DMA32);
> +	if (!virt)
> +		return -ENOMEM;
> +
> +	memset(&request, 0, sizeof(request));
> +	request.addr = phys;
> +
> +	memset(&msg, 0, sizeof(msg));
> +	msg.mrq = MRQ_QUERY_TAG;
> +	msg.tx.data = &request;
> +	msg.tx.size = sizeof(request);
> +
> +	local_irq_save(flags);
> +	err = tegra_bpmp_transfer_atomic(bpmp, &msg);
> +	local_irq_restore(flags);

I believe the local_irq_save/restore can be dropped - they are not
adding any value.

> +
> +	if (err == 0)
> +		strlcpy(tag, virt, size);
> +
> +	dma_free_coherent(bpmp->dev, TEGRA_BPMP_MSG_DATA_SIZE, virt, phys);
> +
> +	return err;
> +}
> +
> +static void tegra_bpmp_channel_signal(struct tegra_bpmp_channel *channel)
> +{
> +	unsigned long flags = channel->ob->flags;
> +
> +	if ((flags & MSG_RING) == 0)
> +		return;
> +
> +	complete(&channel->completion);
> +}
> +
> +static void tegra_bpmp_handle_rx(struct mbox_client *client, void *data)
> +{
> +	struct tegra_bpmp *bpmp = mbox_client_to_bpmp(client);
> +	struct tegra_bpmp_channel *channel;
> +	unsigned int i, count;
> +	unsigned long *busy;
> +
> +	channel = tegra_bpmp_channel_get_rx(bpmp);
> +	count = bpmp->soc->channels.thread.count;
> +	busy = bpmp->threaded.busy;
> +
> +	if (tegra_bpmp_master_acked(channel))
> +		tegra_bpmp_handle_mrq(bpmp, channel->ib->code, channel);
> +
> +	spin_lock(&bpmp->lock);
> +
> +	for_each_set_bit(i, busy, count) {
> +		struct tegra_bpmp_channel *channel;
> +
> +		channel = tegra_bpmp_channel_get_thread(bpmp, i);
> +		if (!channel)
> +			continue;
> +
> +		if (tegra_bpmp_master_acked(channel)) {
> +			tegra_bpmp_channel_signal(channel);
> +			clear_bit(i, busy);
> +		}
> +	}
> +
> +	spin_unlock(&bpmp->lock);
> +}
> +
> +static void tegra_bpmp_ivc_notify(struct tegra_ivc *ivc, void *data)
> +{
> +	struct tegra_bpmp *bpmp = data;
> +	int err;
> +
> +	err = mbox_send_message(bpmp->mbox.channel, NULL);
> +	if (err < 0)
> +		return;
> +
> +	mbox_client_txdone(bpmp->mbox.channel, 0);
> +}
> +
> +static int tegra_bpmp_channel_init(struct tegra_bpmp_channel *channel,
> +				   struct tegra_bpmp *bpmp,
> +				   unsigned int index)
> +{
> +	size_t message, queue;
> +	void __iomem *rx_base;
> +	void __iomem *tx_base;
> +	int err;
> +
> +	channel->ivc = devm_kzalloc(bpmp->dev, sizeof(*channel->ivc),
> +				    GFP_KERNEL);
> +	if (!channel->ivc)
> +		return -ENOMEM;
> +
> +	message = tegra_ivc_align(TEGRA_BPMP_MSG_SIZE);
> +	queue = tegra_ivc_total_queue_size(message);
> +
> +	rx_base = bpmp->rx_base + queue * index;
> +	tx_base = bpmp->tx_base + queue * index;
> +
> +	err = tegra_ivc_init(channel->ivc, bpmp->dev, rx_base, DMA_ERROR_CODE,
> +			     tx_base, DMA_ERROR_CODE, 1, message,
> +			     tegra_bpmp_ivc_notify, bpmp);
> +	if (err < 0) {
> +		dev_err(bpmp->dev, "failed to setup IVC for channel %u: %d\n",
> +			index, err);
> +		return err;
> +	}
> +
> +	/* reset the channel state */
> +	tegra_ivc_reset(channel->ivc);
> +
> +	/* sync the channel state with BPMP */
> +	while (tegra_ivc_notified(channel->ivc))
> +		;
> +
> +	init_completion(&channel->completion);
> +	channel->bpmp = bpmp;
> +
> +	return 0;
> +}
> +
> +static int tegra_bpmp_init_powergates(struct tegra_bpmp *bpmp)
> +{
> +	struct mrq_pg_read_state_response response;
> +	struct mrq_pg_read_state_request request;
> +	struct tegra_bpmp_message msg;
> +	unsigned int i;
> +	int err;
> +
> +	dev_dbg(bpmp->dev, "powergates:\n");
> +
> +	for (i = 0; i < 32; i++) {
> +		memset(&request, 0, sizeof(request));
> +		request.partition_id = i;
> +
> +		memset(&response, 0, sizeof(response));
> +
> +		memset(&msg, 0, sizeof(msg));
> +		msg.mrq = MRQ_PG_READ_STATE;
> +		msg.tx.data = &request;
> +		msg.tx.size = sizeof(request);
> +		msg.rx.data = &response;
> +		msg.rx.size = sizeof(response);
> +
> +		err = tegra_bpmp_transfer(bpmp, &msg);
> +		if (err < 0) {
> +			dev_err(bpmp->dev, "failed to transfer message: %d\n", err);
> +			continue;
> +		}
> +
> +		dev_dbg(bpmp->dev, "  %u: %x (%x)\n", i, response.logic_state,
> +			response.sram_state);
> +	}
> +
> +	return 0;
> +}

IMO, this should not be here (in the firmware interface driver). We
should keep the bpmp driver and its users de-coupled. 

> +
> +static int tegra_bpmp_probe(struct platform_device *pdev)
> +{
> +	struct tegra_bpmp_channel *channel;
> +	struct tegra_bpmp *bpmp;
> +	struct device_node *np;
> +	struct resource res;
> +	unsigned int i;
> +	char tag[32];
> +	size_t size;
> +	int err;
> +
> +	bpmp = devm_kzalloc(&pdev->dev, sizeof(*bpmp), GFP_KERNEL);
> +	if (!bpmp)
> +		return -ENOMEM;
> +
> +	bpmp->soc = of_device_get_match_data(&pdev->dev);
> +	bpmp->dev = &pdev->dev;
> +
> +	np = of_parse_phandle(pdev->dev.of_node, "shmem", 0);
> +	if (!np)
> +		return -ENOENT;
> +
> +	of_address_to_resource(np, 0, &res);
> +	of_node_put(np);
> +
> +	bpmp->tx_base = devm_ioremap_resource(&pdev->dev, &res);
> +	if (IS_ERR(bpmp->tx_base))
> +		return PTR_ERR(bpmp->tx_base);
> +
> +	np = of_parse_phandle(pdev->dev.of_node, "shmem", 1);
> +	if (!np)
> +		return -ENOENT;
> +
> +	of_address_to_resource(np, 0, &res);
> +	of_node_put(np);
> +
> +	bpmp->rx_base = devm_ioremap_resource(&pdev->dev, &res);
> +	if (IS_ERR(bpmp->rx_base))
> +		return PTR_ERR(bpmp->rx_base);
> +
> +	bpmp->num_channels = bpmp->soc->channels.cpu_tx.count +
> +			     bpmp->soc->channels.thread.count +
> +			     bpmp->soc->channels.cpu_rx.count;
> +
> +	bpmp->channels = devm_kcalloc(&pdev->dev, bpmp->num_channels,
> +				      sizeof(*channel), GFP_KERNEL);
> +	if (!bpmp->channels)
> +		return -ENOMEM;
> +
> +	/* mbox registration */
> +	bpmp->mbox.client.dev = &pdev->dev;
> +	bpmp->mbox.client.rx_callback = tegra_bpmp_handle_rx;
> +	bpmp->mbox.client.tx_block = false;
> +	bpmp->mbox.client.knows_txdone = false;
> +
> +	bpmp->mbox.channel = mbox_request_channel(&bpmp->mbox.client, 0);
> +	if (IS_ERR(bpmp->mbox.channel)) {
> +		err = PTR_ERR(bpmp->mbox.channel);
> +		dev_err(&pdev->dev, "failed to get HSP mailbox: %d\n", err);
> +		return err;
> +	}
> +
> +	/* message channel initialization */
> +	for (i = 0; i < bpmp->num_channels; i++) {
> +		struct tegra_bpmp_channel *channel = &bpmp->channels[i];
> +
> +		err = tegra_bpmp_channel_init(channel, bpmp, i);
> +		if (err)
> +			return err;
> +	}
> +
> +	bpmp->threaded.count = bpmp->soc->channels.thread.count;
> +	size = BITS_TO_LONGS(bpmp->threaded.count) * sizeof(long);
> +
> +	bpmp->threaded.allocated = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
> +	if (!bpmp->threaded.allocated)
> +		return -ENOMEM;
> +
> +	bpmp->threaded.busy = devm_kzalloc(&pdev->dev, size, GFP_KERNEL);
> +	if (!bpmp->threaded.busy)
> +		return -ENOMEM;
> +
> +	sema_init(&bpmp->threaded.lock, bpmp->threaded.count);
> +
> +	INIT_LIST_HEAD(&bpmp->mrqs);
> +	spin_lock_init(&bpmp->lock);
> +
> +	err = tegra_bpmp_request_mrq(bpmp, MRQ_PING,
> +				     tegra_bpmp_mrq_handle_ping, bpmp);
> +	if (err < 0)
> +		return err;
> +
> +	err = tegra_bpmp_ping(bpmp);
> +	if (err < 0) {
> +		dev_err(&pdev->dev, "failed to ping BPMP: %d\n", err);
> +		goto free_mrq;
> +	}
> +
> +	err = tegra_bpmp_get_firmware_tag(bpmp, tag, sizeof(tag) - 1);
> +	if (err) {
> +		dev_err(&pdev->dev, "failed to get firmware tag: %d\n", err);
> +		goto free_mrq;
> +	}
> +
> +	dev_info(&pdev->dev, "firmware: %s\n", tag);
> +
> +	err = tegra_bpmp_init_clocks(bpmp);
> +	if (err < 0)
> +		return err;
> +
> +	err = tegra_bpmp_init_resets(bpmp);
> +	if (err < 0)
> +		return err;
> +
> +	err = tegra_bpmp_init_powergates(bpmp);
> +	if (err < 0)
> +		return err;

Given a choice, I would want to keep this out of the bpmp driver
altogether.

> +
> +	platform_set_drvdata(pdev, bpmp);
> +
> +	return 0;
> +
> +free_mrq:
> +	tegra_bpmp_free_mrq(bpmp, MRQ_PING, bpmp);
> +
> +	return err;
> +}
> +
> +static const struct tegra_bpmp_soc tegra186_soc = {
> +	.channels = {
> +		.cpu_tx = {
> +			.offset = 0,
> +			.count = 6,
> +			.timeout = 60 * USEC_PER_SEC,
> +		},
> +		.thread = {
> +			.offset = 6,
> +			.count = 7,
> +			.timeout = 600 * USEC_PER_SEC,
> +		},
> +		.cpu_rx = {
> +			.offset = 13,
> +			.count = 1,
> +			.timeout = 0,
> +		},
> +	},
> +	.num_resets = 193,
> +};
> +
> +static const struct of_device_id tegra_bpmp_match[] = {
> +	{ .compatible = "nvidia,tegra186-bpmp", .data = &tegra186_soc },
> +	{ }
> +};
> +
> +static struct platform_driver tegra_bpmp_driver = {
> +	.driver = {
> +		.name = "tegra-bpmp",
> +		.of_match_table = tegra_bpmp_match,
> +	},
> +	.probe = tegra_bpmp_probe,
> +};
> +
> +static int __init tegra_bpmp_init(void)
> +{
> +	return platform_driver_register(&tegra_bpmp_driver);
> +}
> +core_initcall(tegra_bpmp_init);
> diff --git a/include/soc/tegra/bpmp-abi.h b/include/soc/tegra/bpmp-abi.h
> new file mode 100644
> index 000000000000..0aaef5960e29
> --- /dev/null
> +++ b/include/soc/tegra/bpmp-abi.h
> @@ -0,0 +1,1601 @@
> +/*
> + * Copyright (c) 2014-2016, NVIDIA CORPORATION.  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.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#ifndef _ABI_BPMP_ABI_H_
> +#define _ABI_BPMP_ABI_H_
> +
> +#ifdef LK
> +#include <stdint.h>
> +#endif
> +
> +#ifndef __ABI_PACKED
> +#define __ABI_PACKED __attribute__((packed))
> +#endif
> +
> +#ifdef NO_GCC_EXTENSIONS
> +#define EMPTY char empty;
> +#define EMPTY_ARRAY 1
> +#else
> +#define EMPTY
> +#define EMPTY_ARRAY 0
> +#endif
> +
> +#ifndef __UNION_ANON
> +#define __UNION_ANON
> +#endif
> +/**
> + * @file
> + */
> +
> +
> +/**
> + * @defgroup MRQ MRQ Messages
> + * @brief Messages sent to/from BPMP via IPC
> + * @{
> + *   @defgroup MRQ_Format Message Format
> + *   @defgroup MRQ_Codes Message Request (MRQ) Codes
> + *   @defgroup MRQ_Payloads Message Payloads
> + *   @defgroup Error_Codes Error Codes
> + * @}
> + */
> +
> +/**
> + * @addtogroup MRQ_Format Message Format
> + * @{
> + * The CPU requests the BPMP to perform a particular service by
> + * sending it an IVC frame containing a single MRQ message. An MRQ
> + * message consists of a @ref mrq_request followed by a payload whose
> + * format depends on mrq_request::mrq.
> + *
> + * The BPMP processes the data and replies with an IVC frame (on the
> + * same IVC channel) containing and MRQ response. An MRQ response
> + * consists of a @ref mrq_response followed by a payload whose format
> + * depends on the associated mrq_request::mrq.
> + *
> + * A well-defined subset of the MRQ messages that the CPU sends to the
> + * BPMP can lead to BPMP eventually sending an MRQ message to the
> + * CPU. For example, when the CPU uses an #MRQ_THERMAL message to set
> + * a thermal trip point, the BPMP may eventually send a single
> + * #MRQ_THERMAL message of its own to the CPU indicating that the trip
> + * point has been crossed.
> + * @}
> + */
> +
> +/**
> + * @ingroup MRQ_Format
> + * @brief header for an MRQ message
> + *
> + * Provides the MRQ number for the MRQ message: #mrq. The remainder of
> + * the MRQ message is a payload (immediately following the
> + * mrq_request) whose format depends on mrq.
> + *
> + * @todo document the flags
> + */
> +struct mrq_request {
> +	/** @brief MRQ number of the request */
> +	uint32_t mrq;
> +	/** @brief flags for the request */
> +	uint32_t flags;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Format
> + * @brief header for an MRQ response
> + *
> + *  Provides an error code for the associated MRQ message. The
> + *  remainder of the MRQ response is a payload (immediately following
> + *  the mrq_response) whose format depends on the associated
> + *  mrq_request::mrq
> + *
> + * @todo document the flags
> + */
> +struct mrq_response {
> +	/** @brief error code for the MRQ request itself */
> +	int32_t err;
> +	/** @brief flags for the response */
> +	uint32_t flags;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Format
> + * Minimum needed size for an IPC message buffer
> + */
> +#define MSG_MIN_SZ	128
> +/**
> + * @ingroup MRQ_Format
> + *  Minimum size guaranteed for data in an IPC message buffer
> + */
> +#define MSG_DATA_MIN_SZ	120
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @name Legal MRQ codes
> + * These are the legal values for mrq_request::mrq
> + * @{
> + */
> +
> +#define MRQ_PING		0
> +#define MRQ_QUERY_TAG		1
> +#define MRQ_MODULE_LOAD		4
> +#define MRQ_MODULE_UNLOAD	5
> +#define MRQ_TRACE_MODIFY	7
> +#define MRQ_WRITE_TRACE		8
> +#define MRQ_THREADED_PING	9
> +#define MRQ_MODULE_MAIL		11
> +#define MRQ_DEBUGFS		19
> +#define MRQ_RESET		20
> +#define MRQ_I2C			21
> +#define MRQ_CLK			22
> +#define MRQ_QUERY_ABI		23
> +#define MRQ_PG_READ_STATE	25
> +#define MRQ_PG_UPDATE_STATE	26
> +#define MRQ_THERMAL		27
> +#define MRQ_CPU_VHINT		28
> +#define MRQ_ABI_RATCHET		29
> +#define MRQ_EMC_DVFS_LATENCY	31
> +#define MRQ_TRACE_ITER		64
> +
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @brief Maximum MRQ code to be sent by CPU software to
> + * BPMP. Subject to change in future
> + */
> +#define MAX_CPU_MRQ_ID		64
> +
> +/**
> + * @addtogroup MRQ_Payloads Message Payloads
> + * @{
> + *   @defgroup Ping
> + *   @defgroup Query_Tag Query Tag
> + *   @defgroup Module Loadable Modules
> + *   @defgroup Trace
> + *   @defgroup Debugfs
> + *   @defgroup Reset
> + *   @defgroup I2C
> + *   @defgroup Clocks
> + *   @defgroup ABI_info ABI Info
> + *   @defgroup MC_Flush MC Flush
> + *   @defgroup Powergating
> + *   @defgroup Thermal
> + *   @defgroup Vhint CPU Voltage hint
> + *   @defgroup MRQ_Deprecated Deprecated MRQ messages
> + *   @defgroup EMC
> + * @}
> + */
> +
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_PING
> + * @brief A simple ping
> + *
> + * * Platforms: All
> + * * Initiators: Any
> + * * Targets: Any
> + * * Request Payload: @ref mrq_ping_request
> + * * Response Payload: @ref mrq_ping_response
> + *
> + * @ingroup MRQ_Codes
> + * @def MRQ_THREADED_PING
> + * @brief A deeper ping
> + *
> + * * Platforms: All
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_ping_request
> + * * Response Payload: @ref mrq_ping_response
> + *
> + * Behavior is equivalent to a simple #MRQ_PING except that BPMP
> + * responds from a thread context (providing a slightly more robust
> + * sign of life).
> + *
> + */
> +
> +/**
> + * @ingroup Ping
> + * @brief request with #MRQ_PING
> + *
> + * Used by the sender of an #MRQ_PING message to request a pong from
> + * recipient. The response from the recipient is computed based on
> + * #challenge.
> + */
> +struct mrq_ping_request {
> +/** @brief arbitrarily chosen value */
> +	uint32_t challenge;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Ping
> + * @brief response to #MRQ_PING
> + *
> + * Sent in response to an #MRQ_PING message. #reply should be the
> + * mrq_ping_request challenge left shifted by 1 with the carry-bit
> + * dropped.
> + *
> + */
> +struct mrq_ping_response {
> +	/** @brief response to the MRQ_PING challege */
> +	uint32_t reply;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_QUERY_TAG
> + * @brief Query BPMP firmware's tag (i.e. version information)
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_query_tag_request
> + * * Response Payload: N/A
> + *
> + */
> +
> +/**
> + * @ingroup Query_Tag
> + * @brief request with #MRQ_QUERY_TAG
> + *
> + * Used by #MRQ_QUERY_TAG call to ask BPMP to fill in the memory
> + * pointed by #addr with BPMP firmware header.
> + *
> + * The sender is reponsible for ensuring that #addr is mapped in to
> + * the recipient's address map.
> + */
> +struct mrq_query_tag_request {
> +  /** @brief base address to store the firmware header */
> +	uint32_t addr;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_MODULE_LOAD
> + * @brief dynamically load a BPMP code module
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_module_load_request
> + * * Response Payload: @ref mrq_module_load_response
> + *
> + * @note This MRQ is disabled on production systems
> + *
> + */
> +
> +/**
> + * @ingroup Module
> + * @brief request with #MRQ_MODULE_LOAD
> + *
> + * Used by #MRQ_MODULE_LOAD calls to ask the recipient to dynamically
> + * load the code located at #phys_addr and having size #size
> + * bytes. #phys_addr is treated as a void pointer.
> + *
> + * The recipient copies the code from #phys_addr to locally allocated
> + * memory prior to responding to this message.
> + *
> + * @todo document the module header format
> + *
> + * The sender is responsible for ensuring that the code is mapped in
> + * the recipient's address map.
> + *
> + */
> +struct mrq_module_load_request {
> +	/** @brief base address of the code to load. Treated as (void *) */
> +	uint32_t phys_addr; /* (void *) */
> +	/** @brief size in bytes of code to load */
> +	uint32_t size;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Module
> + * @brief response to #MRQ_MODULE_LOAD
> + *
> + * @todo document mrq_response::err
> + */
> +struct mrq_module_load_response {
> +	/** @brief handle to the loaded module */
> +	uint32_t base;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_MODULE_UNLOAD
> + * @brief unload a previously loaded code module
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_module_unload_request
> + * * Response Payload: N/A
> + *
> + * @note This MRQ is disabled on production systems
> + */
> +
> +/**
> + * @ingroup Module
> + * @brief request with #MRQ_MODULE_UNLOAD
> + *
> + * Used by #MRQ_MODULE_UNLOAD calls to request that a previously loaded
> + * module be unloaded.
> + */
> +struct mrq_module_unload_request {
> +	/** @brief handle of the module to unload */
> +	uint32_t base;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_TRACE_MODIFY
> + * @brief modify the set of enabled trace events
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_trace_modify_request
> + * * Response Payload: @ref mrq_trace_modify_response
> + *
> + * @note This MRQ is disabled on production systems
> + */
> +
> +/**
> + * @ingroup Trace
> + * @brief request with #MRQ_TRACE_MODIFY
> + *
> + * Used by %MRQ_TRACE_MODIFY calls to enable or disable specify trace
> + * events.  #set takes precedence for any bit set in both #set and
> + * #clr.
> + */
> +struct mrq_trace_modify_request {
> +	/** @brief bit mask of trace events to disable */
> +	uint32_t clr;
> +	/** @brief bit mask of trace events to enable */
> +	uint32_t set;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Trace
> + * @brief response to #MRQ_TRACE_MODIFY
> + *
> + * Sent in repsonse to an #MRQ_TRACE_MODIFY message. #mask reflects the
> + * state of which events are enabled after the recipient acted on the
> + * message.
> + *
> + */
> +struct mrq_trace_modify_response {
> +	/** @brief bit mask of trace event enable states */
> +	uint32_t mask;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_WRITE_TRACE
> + * @brief Write trace data to a buffer
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_write_trace_request
> + * * Response Payload: @ref mrq_write_trace_response
> + *
> + * mrq_response::err depends on the @ref mrq_write_trace_request field
> + * values. err is -#BPMP_EINVAL if size is zero or area is NULL or
> + * area is in an illegal range. A positive value for err indicates the
> + * number of bytes written to area.
> + *
> + * @note This MRQ is disabled on production systems
> + */
> +
> +/**
> + * @ingroup Trace
> + * @brief request with #MRQ_WRITE_TRACE
> + *
> + * Used by MRQ_WRITE_TRACE calls to ask the recipient to copy trace
> + * data from the recipient's local buffer to the output buffer. #area
> + * is treated as a byte-aligned pointer in the recipient's address
> + * space.
> + *
> + * The sender is responsible for ensuring that the output
> + * buffer is mapped in the recipient's address map. The recipient is
> + * responsible for protecting its own code and data from accidental
> + * overwrites.
> + */
> +struct mrq_write_trace_request {
> +	/** @brief base address of output buffer */
> +	uint32_t area;
> +	/** @brief size in bytes of the output buffer */
> +	uint32_t size;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Trace
> + * @brief response to #MRQ_WRITE_TRACE
> + *
> + * Once this response is sent, the respondent will not access the
> + * output buffer further.
> + */
> +struct mrq_write_trace_response {
> +	/**
> +	 * @brief flag whether more data remains in local buffer
> +	 *
> +	 * Value is 1 if the entire local trace buffer has been
> +	 * drained to the outputbuffer. Value is 0 otherwise.
> +	 */
> +	uint32_t eof;
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct mrq_threaded_ping_request {
> +	uint32_t challenge;
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct mrq_threaded_ping_response {
> +	uint32_t reply;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_MODULE_MAIL
> + * @brief send a message to a loadable module
> + *
> + * * Platforms: All
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_module_mail_request
> + * * Response Payload: @ref mrq_module_mail_response
> + *
> + * @note This MRQ is disabled on production systems
> + */
> +
> +/**
> + * @ingroup Module
> + * @brief request with #MRQ_MODULE_MAIL
> + */
> +struct mrq_module_mail_request {
> +	/** @brief handle to the previously loaded module */
> +	uint32_t base;
> +	/** @brief module-specific mail payload
> +	 *
> +	 * The length of data[ ] is unknown to the BPMP core firmware
> +	 * but it is limited to the size of an IPC message.
> +	 */
> +	uint8_t data[EMPTY_ARRAY];
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Module
> + * @brief response to #MRQ_MODULE_MAIL
> + */
> +struct mrq_module_mail_response {
> +	/** @brief module-specific mail payload
> +	 *
> +	 * The length of data[ ] is unknown to the BPMP core firmware
> +	 * but it is limited to the size of an IPC message.
> +	 */
> +	uint8_t data[EMPTY_ARRAY];
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_DEBUGFS
> + * @brief Interact with BPMP's debugfs file nodes
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_debugfs_request
> + * * Response Payload: @ref mrq_debugfs_response
> + */
> +
> +/**
> + * @addtogroup Debugfs
> + * @{
> + *
> + * The BPMP firmware implements a pseudo-filesystem called
> + * debugfs. Any driver within the firmware may register with debugfs
> + * to expose an arbitrary set of "files" in the filesystem. When
> + * software on the CPU writes to a debugfs file, debugfs passes the
> + * written data to a callback provided by the driver. When software on
> + * the CPU reads a debugfs file, debugfs queries the driver for the
> + * data to return to the CPU. The intention of the debugfs filesystem
> + * is to provide information useful for debugging the system at
> + * runtime.
> + *
> + * @note The files exposed via debugfs are not part of the
> + * BPMP firmware's ABI. debugfs files may be added or removed in any
> + * given version of the firmware. Typically the semantics of a debugfs
> + * file are consistent from version to version but even that is not
> + * guaranteed.
> + *
> + * @}
> + */
> +/** @ingroup Debugfs */
> +enum mrq_debugfs_commands {
> +	CMD_DEBUGFS_READ = 1,
> +	CMD_DEBUGFS_WRITE = 2,
> +	CMD_DEBUGFS_DUMPDIR = 3,
> +	CMD_DEBUGFS_MAX
> +};
> +
> +/**
> + * @ingroup Debugfs
> + * @brief parameters for CMD_DEBUGFS_READ/WRITE command
> + */
> +struct cmd_debugfs_fileop_request {
> +	/** @brief physical address pointing at filename */
> +	uint32_t fnameaddr;
> +	/** @brief length in bytes of filename buffer */
> +	uint32_t fnamelen;
> +	/** @brief physical address pointing to data buffer */
> +	uint32_t dataaddr;
> +	/** @brief length in bytes of data buffer */
> +	uint32_t datalen;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Debugfs
> + * @brief parameters for CMD_DEBUGFS_READ/WRITE command
> + */
> +struct cmd_debugfs_dumpdir_request {
> +	/** @brief physical address pointing to data buffer */
> +	uint32_t dataaddr;
> +	/** @brief length in bytes of data buffer */
> +	uint32_t datalen;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Debugfs
> + * @brief response data for CMD_DEBUGFS_READ/WRITE command
> + */
> +struct cmd_debugfs_fileop_response {
> +	/** @brief always 0 */
> +	uint32_t reserved;
> +	/** @brief number of bytes read from or written to data buffer */
> +	uint32_t nbytes;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Debugfs
> + * @brief response data for CMD_DEBUGFS_DUMPDIR command
> + */
> +struct cmd_debugfs_dumpdir_response {
> +	/** @brief always 0 */
> +	uint32_t reserved;
> +	/** @brief number of bytes read from or written to data buffer */
> +	uint32_t nbytes;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Debugfs
> + * @brief request with #MRQ_DEBUGFS.
> + *
> + * The sender of an MRQ_DEBUGFS message uses #cmd to specify a debugfs
> + * command to execute. Legal commands are the values of @ref
> + * mrq_debugfs_commands. Each command requires a specific additional
> + * payload of data.
> + *
> + * |command            |payload|
> + * |-------------------|-------|
> + * |CMD_DEBUGFS_READ   |fop    |
> + * |CMD_DEBUGFS_WRITE  |fop    |
> + * |CMD_DEBUGFS_DUMPDIR|dumpdir|
> + */
> +struct mrq_debugfs_request {
> +	uint32_t cmd;
> +	union {
> +		struct cmd_debugfs_fileop_request fop;
> +		struct cmd_debugfs_dumpdir_request dumpdir;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Debugfs
> + */
> +struct mrq_debugfs_response {
> +	/** @brief always 0 */
> +	int32_t reserved;
> +	union {
> +		/** @brief response data for CMD_DEBUGFS_READ OR
> +		 * CMD_DEBUGFS_WRITE command
> +		 */
> +		struct cmd_debugfs_fileop_response fop;
> +		/** @brief response data for CMD_DEBUGFS_DUMPDIR command */
> +		struct cmd_debugfs_dumpdir_response dumpdir;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/**
> + * @addtogroup Debugfs
> + * @{
> + */
> +#define DEBUGFS_S_ISDIR	(1 << 9)
> +#define DEBUGFS_S_IRUSR	(1 << 8)
> +#define DEBUGFS_S_IWUSR	(1 << 7)
> +/** @} */
> +
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_RESET
> + * @brief reset an IP block
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_reset_request
> + * * Response Payload: N/A
> + */
> +
> +/**
> + * @ingroup Reset
> + */
> +enum mrq_reset_commands {
> +	CMD_RESET_ASSERT = 1,
> +	CMD_RESET_DEASSERT = 2,
> +	CMD_RESET_MODULE = 3,
> +	CMD_RESET_MAX, /* not part of ABI and subject to change */
> +};
> +
> +/**
> + * @ingroup Reset
> + * @brief request with MRQ_RESET
> + *
> + * Used by the sender of an #MRQ_RESET message to request BPMP to
> + * assert or or deassert a given reset line.
> + */
> +struct mrq_reset_request {
> +	/** @brief reset action to perform (@enum mrq_reset_commands) */
> +	uint32_t cmd;
> +	/** @brief id of the reset to affected */
> +	uint32_t reset_id;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_I2C
> + * @brief issue an i2c transaction
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_i2c_request
> + * * Response Payload: @ref mrq_i2c_response
> + */
> +
> +/**
> + * @addtogroup I2C
> + * @{
> + */
> +#define TEGRA_I2C_IPC_MAX_IN_BUF_SIZE	(MSG_DATA_MIN_SZ - 12)
> +#define TEGRA_I2C_IPC_MAX_OUT_BUF_SIZE	(MSG_DATA_MIN_SZ - 4)
> +/** @} */
> +
> +/**
> + * @ingroup I2C
> + * @name Serial I2C flags
> + * Use these flags with serial_i2c_request::flags
> + * @{
> + */
> +#define SERIALI2C_TEN           0x0010
> +#define SERIALI2C_RD            0x0001
> +#define SERIALI2C_STOP          0x8000
> +#define SERIALI2C_NOSTART       0x4000
> +#define SERIALI2C_REV_DIR_ADDR  0x2000
> +#define SERIALI2C_IGNORE_NAK    0x1000
> +#define SERIALI2C_NO_RD_ACK     0x0800
> +#define SERIALI2C_RECV_LEN      0x0400
> +/** @} */
> +/** @ingroup I2C */
> +enum {
> +	CMD_I2C_XFER = 1
> +};
> +
> +/**
> + * @ingroup I2C
> + * @brief serializable i2c request
> + *
> + * Instances of this structure are packed (little-endian) into
> + * cmd_i2c_xfer_request::data_buf. Each instance represents a single
> + * transaction (or a portion of a transaction with repeated starts) on
> + * an i2c bus.
> + *
> + * Because these structures are packed, some instances are likely to
> + * be misaligned. Additionally because #data is variable length, it is
> + * not possible to iterate through a serialized list of these
> + * structures without inspecting #len in each instance.  It may be
> + * easier to serialize or deserialize cmd_i2c_xfer_request::data_buf
> + * manually rather than using this structure definition.
> +*/
> +struct serial_i2c_request {
> +	/** @brief I2C slave address */
> +	uint16_t addr;
> +	/** @brief bitmask of SERIALI2C_ flags */
> +	uint16_t flags;
> +	/** @brief length of I2C transaction in bytes */
> +	uint16_t len;
> +	/** @brief for write transactions only, #len bytes of data */
> +	uint8_t data[];
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup I2C
> + * @brief trigger one or more i2c transactions
> + */
> +struct cmd_i2c_xfer_request {
> +	/** @brief valid bus number from mach-t186/i2c-t186.h*/
> +	uint32_t bus_id;
> +
> +	/** @brief count of valid bytes in #data_buf*/
> +	uint32_t data_size;
> +
> +	/** @brief serialized packed instances of @ref serial_i2c_request*/
> +	uint8_t data_buf[TEGRA_I2C_IPC_MAX_IN_BUF_SIZE];
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup I2C
> + * @brief container for data read from the i2c bus
> + *
> + * Processing an cmd_i2c_xfer_request::data_buf causes BPMP to execute
> + * zero or more I2C reads. The data read from the bus is serialized
> + * into #data_buf.
> + */
> +struct cmd_i2c_xfer_response {
> +	/** @brief count of valid bytes in #data_buf*/
> +	uint32_t data_size;
> +	/** @brief i2c read data */
> +	uint8_t data_buf[TEGRA_I2C_IPC_MAX_OUT_BUF_SIZE];
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup I2C
> + * @brief request with #MRQ_I2C
> + */
> +struct mrq_i2c_request {
> +	/** @brief always CMD_I2C_XFER (i.e. 1) */
> +	uint32_t cmd;
> +	/** @brief parameters of the transfer request */
> +	struct cmd_i2c_xfer_request xfer;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup I2C
> + * @brief response to #MRQ_I2C
> + */
> +struct mrq_i2c_response {
> +	struct cmd_i2c_xfer_response xfer;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_CLK
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_clk_request
> + * * Response Payload: @ref mrq_clk_response
> + * @addtogroup Clocks
> + * @{
> + */
> +
> +/**
> + * @name MRQ_CLK sub-commands
> + * @{
> + */
> +enum {
> +	CMD_CLK_GET_RATE = 1,
> +	CMD_CLK_SET_RATE = 2,
> +	CMD_CLK_ROUND_RATE = 3,
> +	CMD_CLK_GET_PARENT = 4,
> +	CMD_CLK_SET_PARENT = 5,
> +	CMD_CLK_IS_ENABLED = 6,
> +	CMD_CLK_ENABLE = 7,
> +	CMD_CLK_DISABLE = 8,
> +	CMD_CLK_GET_ALL_INFO = 14,
> +	CMD_CLK_GET_MAX_CLK_ID = 15,
> +	CMD_CLK_MAX,
> +};
> +/** @} */
> +
> +#define MRQ_CLK_NAME_MAXLEN	40
> +#define MRQ_CLK_MAX_PARENTS	16
> +
> +/** @private */
> +struct cmd_clk_get_rate_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +struct cmd_clk_get_rate_response {
> +	int64_t rate;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_set_rate_request {
> +	int32_t unused;
> +	int64_t rate;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_set_rate_response {
> +	int64_t rate;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_round_rate_request {
> +	int32_t unused;
> +	int64_t rate;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_round_rate_response {
> +	int64_t rate;
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_get_parent_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +struct cmd_clk_get_parent_response {
> +	uint32_t parent_id;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_set_parent_request {
> +	uint32_t parent_id;
> +} __ABI_PACKED;
> +
> +struct cmd_clk_set_parent_response {
> +	uint32_t parent_id;
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_is_enabled_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +struct cmd_clk_is_enabled_response {
> +	int32_t state;
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_enable_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_enable_response {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_disable_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_disable_response {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_get_all_info_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +struct cmd_clk_get_all_info_response {
> +	uint32_t flags;
> +	uint32_t parent;
> +	uint32_t parents[MRQ_CLK_MAX_PARENTS];
> +	uint8_t num_parents;
> +	uint8_t name[MRQ_CLK_NAME_MAXLEN];
> +} __ABI_PACKED;
> +
> +/** @private */
> +struct cmd_clk_get_max_clk_id_request {
> +	EMPTY
> +} __ABI_PACKED;
> +
> +struct cmd_clk_get_max_clk_id_response {
> +	uint32_t max_id;
> +} __ABI_PACKED;
> +/** @} */
> +
> +/**
> + * @ingroup Clocks
> + * @brief request with #MRQ_CLK
> + *
> + * Used by the sender of an #MRQ_CLK message to control clocks. The
> + * clk_request is split into several sub-commands. Some sub-commands
> + * require no additional data. Others have a sub-command specific
> + * payload
> + *
> + * |sub-command                 |payload                |
> + * |----------------------------|-----------------------|
> + * |CMD_CLK_GET_RATE            |-                      |
> + * |CMD_CLK_SET_RATE            |clk_set_rate           |
> + * |CMD_CLK_ROUND_RATE          |clk_round_rate         |
> + * |CMD_CLK_GET_PARENT          |-                      |
> + * |CMD_CLK_SET_PARENT          |clk_set_parent         |
> + * |CMD_CLK_IS_ENABLED          |-                      |
> + * |CMD_CLK_ENABLE              |-                      |
> + * |CMD_CLK_DISABLE             |-                      |
> + * |CMD_CLK_GET_ALL_INFO        |-                      |
> + * |CMD_CLK_GET_MAX_CLK_ID      |-                      |
> + *
> + */
> +
> +struct mrq_clk_request {
> +	/** @brief sub-command and clock id concatenated to 32-bit word.
> +	 * - bits[31..24] is the sub-cmd.
> +	 * - bits[23..0] is the clock id
> +	 */
> +	uint32_t cmd_and_id;
> +
> +	union {
> +		/** @private */
> +		struct cmd_clk_get_rate_request clk_get_rate;
> +		struct cmd_clk_set_rate_request clk_set_rate;
> +		struct cmd_clk_round_rate_request clk_round_rate;
> +		/** @private */
> +		struct cmd_clk_get_parent_request clk_get_parent;
> +		struct cmd_clk_set_parent_request clk_set_parent;
> +		/** @private */
> +		struct cmd_clk_enable_request clk_enable;
> +		/** @private */
> +		struct cmd_clk_disable_request clk_disable;
> +		/** @private */
> +		struct cmd_clk_is_enabled_request clk_is_enabled;
> +		/** @private */
> +		struct cmd_clk_get_all_info_request clk_get_all_info;
> +		/** @private */
> +		struct cmd_clk_get_max_clk_id_request clk_get_max_clk_id;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup Clocks
> + * @brief response to MRQ_CLK
> + *
> + * Each sub-command supported by @ref mrq_clk_request may return
> + * sub-command-specific data. Some do and some do not as indicated in
> + * the following table
> + *
> + * |sub-command                 |payload                 |
> + * |----------------------------|------------------------|
> + * |CMD_CLK_GET_RATE            |clk_get_rate            |
> + * |CMD_CLK_SET_RATE            |clk_set_rate            |
> + * |CMD_CLK_ROUND_RATE          |clk_round_rate          |
> + * |CMD_CLK_GET_PARENT          |clk_get_parent          |
> + * |CMD_CLK_SET_PARENT          |clk_set_parent          |
> + * |CMD_CLK_IS_ENABLED          |clk_is_enabled          |
> + * |CMD_CLK_ENABLE              |-                       |
> + * |CMD_CLK_DISABLE             |-                       |
> + * |CMD_CLK_GET_ALL_INFO        |clk_get_all_info        |
> + * |CMD_CLK_GET_MAX_CLK_ID      |clk_get_max_id          |
> + *
> + */
> +
> +struct mrq_clk_response {
> +	union {
> +		struct cmd_clk_get_rate_response clk_get_rate;
> +		struct cmd_clk_set_rate_response clk_set_rate;
> +		struct cmd_clk_round_rate_response clk_round_rate;
> +		struct cmd_clk_get_parent_response clk_get_parent;
> +		struct cmd_clk_set_parent_response clk_set_parent;
> +		/** @private */
> +		struct cmd_clk_enable_response clk_enable;
> +		/** @private */
> +		struct cmd_clk_disable_response clk_disable;
> +		struct cmd_clk_is_enabled_response clk_is_enabled;
> +		struct cmd_clk_get_all_info_response clk_get_all_info;
> +		struct cmd_clk_get_max_clk_id_response clk_get_max_clk_id;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_QUERY_ABI
> + * @brief check if an MRQ is implemented
> + *
> + * * Platforms: All
> + * * Initiators: Any
> + * * Targets: Any
> + * * Request Payload: @ref mrq_query_abi_request
> + * * Response Payload: @ref mrq_query_abi_response
> + */
> +
> +/**
> + * @ingroup ABI_info
> + * @brief request with MRQ_QUERY_ABI
> + *
> + * Used by #MRQ_QUERY_ABI call to check if MRQ code #mrq is supported
> + * by the recipient.
> + */
> +struct mrq_query_abi_request {
> +	/** @brief MRQ code to query */
> +	uint32_t mrq;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup ABI_info
> + * @brief response to MRQ_QUERY_ABI
> + */
> +struct mrq_query_abi_response {
> +	/** @brief 0 if queried MRQ is supported. Else, -#BPMP_ENODEV */
> +	int32_t status;
> +} __ABI_PACKED;
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_PG_READ_STATE
> + * @brief read the power-gating state of a partition
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_pg_read_state_request
> + * * Response Payload: @ref mrq_pg_read_state_response
> + * @addtogroup Powergating
> + * @{
> + */
> +
> +/**
> + * @brief request with #MRQ_PG_READ_STATE
> + *
> + * Used by MRQ_PG_READ_STATE call to read the current state of a
> + * partition.
> + */
> +struct mrq_pg_read_state_request {
> +	/** @brief ID of partition */
> +	uint32_t partition_id;
> +} __ABI_PACKED;
> +
> +/**
> + * @brief response to MRQ_PG_READ_STATE
> + * @todo define possible errors.
> + */
> +struct mrq_pg_read_state_response {
> +	/** @brief read as don't care */
> +	uint32_t sram_state;
> +	/** @brief state of power partition
> +	 * * 0 : off
> +	 * * 1 : on
> +	 */
> +	uint32_t logic_state;
> +} __ABI_PACKED;
> +
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_PG_UPDATE_STATE
> + * @brief modify the power-gating state of a partition
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_pg_update_state_request
> + * * Response Payload: N/A
> + * @addtogroup Powergating
> + * @{
> + */
> +
> +/**
> + * @brief request with mrq_pg_update_state_request
> + *
> + * Used by #MRQ_PG_UPDATE_STATE call to request BPMP to change the
> + * state of a power partition #partition_id.
> + */
> +struct mrq_pg_update_state_request {
> +	/** @brief ID of partition */
> +	uint32_t partition_id;
> +	/** @brief secondary control of power partition
> +	 *  @details Ignored by many versions of the BPMP
> +	 *  firmware. For maximum compatibility, set the value
> +	 *  according to @logic_state
> +	 * *  0x1: power ON partition (@ref logic_state == 0x3)
> +	 * *  0x3: power OFF partition (@ref logic_state == 0x1)
> +	 */
> +	uint32_t sram_state;
> +	/** @brief controls state of power partition, legal values are
> +	 * *  0x1 : power OFF partition
> +	 * *  0x3 : power ON partition
> +	 */
> +	uint32_t logic_state;
> +	/** @brief change state of clocks of the power partition, legal values
> +	 * *  0x0 : do not change clock state
> +	 * *  0x1 : disable partition clocks (only applicable when
> +	 *          @ref logic_state == 0x1)
> +	 * *  0x3 : enable partition clocks (only applicable when
> +	 *          @ref logic_state == 0x3)
> +	 */
> +	uint32_t clock_state;
> +} __ABI_PACKED;
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_THERMAL
> + * @brief interact with BPMP thermal framework
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: Any
> + * * Request Payload: TODO
> + * * Response Payload: TODO
> + *
> + * @addtogroup Thermal
> + *
> + * The BPMP firmware includes a thermal framework. Drivers within the
> + * bpmp firmware register with the framework to provide thermal
> + * zones. Each thermal zone corresponds to an entity whose temperature
> + * can be measured. The framework also has a notion of trip points. A
> + * trip point consists of a thermal zone id, a temperature, and a
> + * callback routine. The framework invokes the callback when the zone
> + * hits the indicated temperature. The BPMP firmware uses this thermal
> + * framework interally to implement various temperature-dependent
> + * functions.
> + *
> + * Software on the CPU can use #MRQ_THERMAL (with payload @ref
> + * mrq_thermal_host_to_bpmp_request) to interact with the BPMP thermal
> + * framework. The CPU must It can query the number of supported zones,
> + * query zone temperatures, and set trip points.
> + *
> + * When a trip point set by the CPU gets crossed, BPMP firmware issues
> + * an IPC to the CPU having mrq_request::mrq = #MRQ_THERMAL and a
> + * payload of @ref mrq_thermal_bpmp_to_host_request.
> + * @{
> + */
> +enum mrq_thermal_host_to_bpmp_cmd {
> +	/**
> +	 * @brief Check whether the BPMP driver supports the specified
> +	 * request type.
> +	 *
> +	 * Host needs to supply request parameters.
> +	 *
> +	 * mrq_response::err is 0 if the specified request is
> +	 * supported and -#BPMP_ENODEV otherwise.
> +	 */
> +	CMD_THERMAL_QUERY_ABI = 0,
> +
> +	/**
> +	 * @brief Get the current temperature of the specified zone.
> +	 *
> +	 * Host needs to supply request parameters.
> +	 *
> +	 * mrq_response::err is
> +	 * *  0: Temperature query succeeded.
> +	 * *  -#BPMP_EINVAL: Invalid request parameters.
> +	 * *  -#BPMP_ENOENT: No driver registered for thermal zone..
> +	 * *  -#BPMP_EFAULT: Problem reading temperature measurement.
> +	 */
> +	CMD_THERMAL_GET_TEMP = 1,
> +
> +	/**
> +	 * @brief Enable or disable and set the lower and upper
> +	 *   thermal limits for a thermal trip point. Each zone has
> +	 *   one trip point.
> +	 *
> +	 * Host needs to supply request parameters. Once the
> +	 * temperature hits a trip point, the BPMP will send a message
> +	 * to the CPU having MRQ=MRQ_THERMAL and
> +	 * type=CMD_THERMAL_HOST_TRIP_REACHED
> +	 *
> +	 * mrq_response::err is
> +	 * *  0: Trip successfully set.
> +	 * *  -#BPMP_EINVAL: Invalid request parameters.
> +	 * *  -#BPMP_ENOENT: No driver registered for thermal zone.
> +	 * *  -#BPMP_EFAULT: Problem setting trip point.
> +	 */
> +	CMD_THERMAL_SET_TRIP = 2,
> +
> +	/**
> +	 * @brief Get the number of supported thermal zones.
> +	 *
> +	 * No request parameters required.
> +	 *
> +	 * mrq_response::err is always 0, indicating success.
> +	 */
> +	CMD_THERMAL_GET_NUM_ZONES = 3,
> +
> +	/** @brief: number of supported host-to-bpmp commands. May
> +	 * increase in future
> +	 */
> +	CMD_THERMAL_HOST_TO_BPMP_NUM
> +};
> +
> +enum mrq_thermal_bpmp_to_host_cmd {
> +	/**
> +	 * @brief Indication that the temperature for a zone has
> +	 *   exceeded the range indicated in the thermal trip point
> +	 *   for the zone.
> +	 *
> +	 * BPMP needs to supply request parameters. Host only needs to
> +	 * acknowledge.
> +	 */
> +	CMD_THERMAL_HOST_TRIP_REACHED = 100,
> +
> +	/** @brief: number of supported bpmp-to-host commands. May
> +	 * increase in future
> +	 */
> +	CMD_THERMAL_BPMP_TO_HOST_NUM
> +};
> +
> +/*
> + * Host->BPMP request data for request type CMD_THERMAL_QUERY_ABI
> + *
> + * zone: Request type for which to check existence.
> + */
> +struct cmd_thermal_query_abi_request {
> +	uint32_t type;
> +} __ABI_PACKED;
> +
> +/*
> + * Host->BPMP request data for request type CMD_THERMAL_GET_TEMP
> + *
> + * zone: Number of thermal zone.
> + */
> +struct cmd_thermal_get_temp_request {
> +	uint32_t zone;
> +} __ABI_PACKED;
> +
> +/*
> + * BPMP->Host reply data for request CMD_THERMAL_GET_TEMP
> + *
> + * error: 0 if request succeeded.
> + *	-BPMP_EINVAL if request parameters were invalid.
> + *      -BPMP_ENOENT if no driver was registered for the specified thermal zone.
> + *      -BPMP_EFAULT for other thermal zone driver errors.
> + * temp: Current temperature in millicelsius.
> + */
> +struct cmd_thermal_get_temp_response {
> +	int32_t temp;
> +} __ABI_PACKED;
> +
> +/*
> + * Host->BPMP request data for request type CMD_THERMAL_SET_TRIP
> + *
> + * zone: Number of thermal zone.
> + * low: Temperature of lower trip point in millicelsius
> + * high: Temperature of upper trip point in millicelsius
> + * enabled: 1 to enable trip point, 0 to disable trip point
> + */
> +struct cmd_thermal_set_trip_request {
> +	uint32_t zone;
> +	int32_t low;
> +	int32_t high;
> +	uint32_t enabled;
> +} __ABI_PACKED;
> +
> +/*
> + * BPMP->Host request data for request type CMD_THERMAL_HOST_TRIP_REACHED
> + *
> + * zone: Number of thermal zone where trip point was reached.
> + */
> +struct cmd_thermal_host_trip_reached_request {
> +	uint32_t zone;
> +} __ABI_PACKED;
> +
> +/*
> + * BPMP->Host reply data for request type CMD_THERMAL_GET_NUM_ZONES
> + *
> + * num: Number of supported thermal zones. The thermal zones are indexed
> + *      starting from zero.
> + */
> +struct cmd_thermal_get_num_zones_response {
> +	uint32_t num;
> +} __ABI_PACKED;
> +
> +/*
> + * Host->BPMP request data.
> + *
> + * Reply type is union mrq_thermal_bpmp_to_host_response.
> + *
> + * type: Type of request. Values listed in enum mrq_thermal_type.
> + * data: Request type specific parameters.
> + */
> +struct mrq_thermal_host_to_bpmp_request {
> +	uint32_t type;
> +	union {
> +		struct cmd_thermal_query_abi_request query_abi;
> +		struct cmd_thermal_get_temp_request get_temp;
> +		struct cmd_thermal_set_trip_request set_trip;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/*
> + * BPMP->Host request data.
> + *
> + * type: Type of request. Values listed in enum mrq_thermal_type.
> + * data: Request type specific parameters.
> + */
> +struct mrq_thermal_bpmp_to_host_request {
> +	uint32_t type;
> +	union {
> +		struct cmd_thermal_host_trip_reached_request host_trip_reached;
> +	} __UNION_ANON;
> +} __ABI_PACKED;
> +
> +/*
> + * Data in reply to a Host->BPMP request.
> + */
> +union mrq_thermal_bpmp_to_host_response {
> +	struct cmd_thermal_get_temp_response get_temp;
> +	struct cmd_thermal_get_num_zones_response get_num_zones;
> +} __ABI_PACKED;
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_CPU_VHINT
> + * @brief Query CPU voltage hint data
> + *
> + * * Platforms: T186
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_cpu_vhint_request
> + * * Response Payload: N/A
> + *
> + * @addtogroup Vhint CPU Voltage hint
> + * @{
> + */
> +
> +/**
> + * @brief request with #MRQ_CPU_VHINT
> + *
> + * Used by #MRQ_CPU_VHINT call by CCPLEX to retrieve voltage hint data
> + * from BPMP to memory space pointed by #addr. CCPLEX is responsible
> + * to allocate sizeof(cpu_vhint_data) sized block of memory and
> + * appropriately map it for BPMP before sending the request.
> + */
> +struct mrq_cpu_vhint_request {
> +	/** @brief IOVA address for the #cpu_vhint_data */
> +	uint32_t addr; /* struct cpu_vhint_data * */
> +	/** @brief ID of the cluster whose data is requested */
> +	uint32_t cluster_id; /* enum cluster_id */
> +} __ABI_PACKED;
> +
> +/**
> + * @brief description of the CPU v/f relation
> + *
> + * Used by #MRQ_CPU_VHINT call to carry data pointed by #addr of
> + * struct mrq_cpu_vhint_request
> + */
> +struct cpu_vhint_data {
> +	uint32_t ref_clk_hz; /**< reference frequency in Hz */
> +	uint16_t pdiv; /**< post divider value */
> +	uint16_t mdiv; /**< input divider value */
> +	uint16_t ndiv_max; /**< fMAX expressed with max NDIV value */
> +	/** table of ndiv values as a function of vINDEX (voltage index) */
> +	uint16_t ndiv[80];
> +	/** minimum allowed NDIV value */
> +	uint16_t ndiv_min;
> +	/** minimum allowed voltage hint value (as in vINDEX) */
> +	uint16_t vfloor;
> +	/** maximum allowed voltage hint value (as in vINDEX) */
> +	uint16_t vceil;
> +	/** post-multiplier for vindex value */
> +	uint16_t vindex_mult;
> +	/** post-divider for vindex value */
> +	uint16_t vindex_div;
> +	/** reserved for future use */
> +	uint16_t reserved[328];
> +} __ABI_PACKED;
> +
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_ABI_RATCHET
> + * @brief ABI ratchet value query
> + *
> + * * Platforms: T186
> + * * Initiators: Any
> + * * Targets: BPMP
> + * * Request Payload: @ref mrq_abi_ratchet_request
> + * * Response Payload: @ref mrq_abi_ratchet_response
> + * @addtogroup ABI_info
> + * @{
> + */
> +
> +/**
> + * @brief an ABI compatibility mechanism
> + *
> + * BPMP_ABI_RATCHET_VALUE may increase for various reasons in a future
> + * revision of this header file.
> + * 1. That future revision deprecates some MRQ
> + * 2. That future revision introduces a breaking change to an existing
> + *    MRQ or
> + * 3. A bug is discovered in an existing implementation of the BPMP-FW
> + *    (or possibly one of its clients) which warrants deprecating that
> + *    implementation.
> + */
> +#define BPMP_ABI_RATCHET_VALUE 3
> +
> +/**
> + * @brief request with #MRQ_ABI_RATCHET.
> + *
> + * #ratchet should be #BPMP_ABI_RATCHET_VALUE from the ABI header
> + * against which the requester was compiled.
> + *
> + * If ratchet is less than BPMP's #BPMP_ABI_RATCHET_VALUE, BPMP may
> + * reply with mrq_response::err = -#BPMP_ERANGE to indicate that
> + * BPMP-FW cannot interoperate correctly with the requester. Requester
> + * should cease further communication with BPMP.
> + *
> + * Otherwise, err shall be 0.
> + */
> +struct mrq_abi_ratchet_request {
> +	/** @brief requester's ratchet value */
> +	uint16_t ratchet;
> +};
> +
> +/**
> + * @brief response to #MRQ_ABI_RATCHET
> + *
> + * #ratchet shall be #BPMP_ABI_RATCHET_VALUE from the ABI header
> + * against which BPMP firwmare was compiled.
> + *
> + * If #ratchet is less than the requester's #BPMP_ABI_RATCHET_VALUE,
> + * the requster must either interoperate with BPMP according to an ABI
> + * header version with BPMP_ABI_RATCHET_VALUE = ratchet or cease
> + * communication with BPMP.
> + *
> + * If mrq_response::err is 0 and ratchet is greater than or equal to the
> + * requester's BPMP_ABI_RATCHET_VALUE, the requester should continue
> + * normal operation.
> + */
> +struct mrq_abi_ratchet_response {
> +	/** @brief BPMP's ratchet value */
> +	uint16_t ratchet;
> +};
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_EMC_DVFS_LATENCY
> + * @brief query frequency dependent EMC DVFS latency
> + *
> + * * Platforms: T186
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: N/A
> + * * Response Payload: @ref mrq_emc_dvfs_latency_response
> + * @addtogroup EMC
> + * @{
> + */
> +
> +/**
> + * @brief used by @ref mrq_emc_dvfs_latency_response
> + */
> +struct emc_dvfs_latency {
> +	/** @brief EMC frequency in kHz */
> +	uint32_t freq;
> +	/** @brief EMC DVFS latency in nanoseconds */
> +	uint32_t latency;
> +} __ABI_PACKED;
> +
> +#define EMC_DVFS_LATENCY_MAX_SIZE	14
> +/**
> + * @brief response to #MRQ_EMC_DVFS_LATENCY
> + */
> +struct mrq_emc_dvfs_latency_response {
> +	/** @brief the number valid entries in #pairs */
> +	uint32_t num_pairs;
> +	/** @brief EMC <frequency, latency> information */
> +	struct emc_dvfs_latency pairs[EMC_DVFS_LATENCY_MAX_SIZE];
> +} __ABI_PACKED;
> +
> +/** @} */
> +
> +/**
> + * @ingroup MRQ_Codes
> + * @def MRQ_TRACE_ITER
> + * @brief manage the trace iterator
> + *
> + * * Platforms: All
> + * * Initiators: CCPLEX
> + * * Targets: BPMP
> + * * Request Payload: N/A
> + * * Response Payload: @ref mrq_trace_iter_request
> + * @addtogroup Trace
> + * @{
> + */
> +enum {
> +	/** @brief (re)start the tracing now. Ignore older events */
> +	TRACE_ITER_INIT = 0,
> +	/** @brief clobber all events in the trace buffer */
> +	TRACE_ITER_CLEAN = 1
> +};
> +
> +/**
> + * @brief request with #MRQ_TRACE_ITER
> + */
> +struct mrq_trace_iter_request {
> +	/** @brief TRACE_ITER_INIT or TRACE_ITER_CLEAN */
> +	uint32_t cmd;
> +} __ABI_PACKED;
> +
> +/** @} */
> +
> +/*
> + *  4. Enumerations
> + */
> +
> +/*
> + *   4.1 CPU enumerations
> + *
> + * See <mach-t186/system-t186.h>
> + *
> + *   4.2 CPU Cluster enumerations
> + *
> + * See <mach-t186/system-t186.h>
> + *
> + *   4.3 System low power state enumerations
> + *
> + * See <mach-t186/system-t186.h>
> + */
> +
> +/*
> + *   4.4 Clock enumerations
> + *
> + * For clock enumerations, see <mach-t186/clk-t186.h>
> + */
> +
> +/*
> + *   4.5 Reset enumerations
> + *
> + * For reset enumerations, see <mach-t186/reset-t186.h>
> + */
> +
> +/*
> + *   4.6 Thermal sensor enumerations
> + *
> + * For thermal sensor enumerations, see <mach-t186/thermal-t186.h>
> + */
> +
> +/**
> + * @defgroup Error_Codes
> + * Negative values for mrq_response::err generally indicate some
> + * error. The ABI defines the following error codes. Negating these
> + * defines is an exercise left to the user.
> + * @{
> + */
> +/** @brief No such file or directory */
> +#define BPMP_ENOENT	2
> +/** @brief No MRQ handler */
> +#define BPMP_ENOHANDLER	3
> +/** @brief I/O error */
> +#define BPMP_EIO	5
> +/** @brief Bad sub-MRQ command */
> +#define BPMP_EBADCMD	6
> +/** @brief Not enough memory */
> +#define BPMP_ENOMEM	12
> +/** @brief Permission denied */
> +#define BPMP_EACCES	13
> +/** @brief Bad address */
> +#define BPMP_EFAULT	14
> +/** @brief No such device */
> +#define BPMP_ENODEV	19
> +/** @brief Argument is a directory */
> +#define BPMP_EISDIR	21
> +/** @brief Invalid argument */
> +#define BPMP_EINVAL	22
> +/** @brief Timeout during operation */
> +#define BPMP_ETIMEDOUT  23
> +/** @brief Out of range */
> +#define BPMP_ERANGE	34
> +/** @} */
> +/** @} */
> +#endif
> diff --git a/include/soc/tegra/bpmp.h b/include/soc/tegra/bpmp.h
> new file mode 100644
> index 000000000000..e17d091e8a9a
> --- /dev/null
> +++ b/include/soc/tegra/bpmp.h
> @@ -0,0 +1,122 @@
> +/*
> + * Copyright (c) 2016, NVIDIA CORPORATION.  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.
> + */
> +
> +#ifndef __SOC_TEGRA_BPMP_H
> +#define __SOC_TEGRA_BPMP_H
> +
> +#include <linux/mailbox_client.h>
> +#include <linux/reset-controller.h>
> +#include <linux/semaphore.h>
> +#include <linux/types.h>
> +
> +#define TEGRA_BPMP_MSG_SIZE 128
> +#define TEGRA_BPMP_MSG_DATA_SIZE 120

Why aren't we using the ABI equivalents?

> +
> +struct tegra_bpmp_soc {
> +	struct {
> +		struct {
> +			unsigned int offset;
> +			unsigned int count;
> +			unsigned int timeout;
> +		} cpu_tx, thread, cpu_rx;
> +	} channels;
> +	unsigned int num_resets;
> +};
> +
> +struct tegra_bpmp_mb_data {
> +	u32 code;
> +	u32 flags;
> +	u8 data[TEGRA_BPMP_MSG_DATA_SIZE];
> +} __packed;
> +
> +struct tegra_bpmp_channel {
> +	struct tegra_bpmp *bpmp;
> +	struct tegra_bpmp_mb_data *ib;
> +	struct tegra_bpmp_mb_data *ob;
> +	struct completion completion;
> +	struct tegra_ivc *ivc;
> +};
> +
> +typedef void (*tegra_bpmp_mrq_handler_t)(unsigned int mrq,
> +					 struct tegra_bpmp_channel *channel,
> +					 void *data);
> +
> +struct tegra_bpmp_mrq {
> +	struct list_head list;
> +	unsigned int mrq;
> +	tegra_bpmp_mrq_handler_t handler;
> +	void *data;
> +};
> +
> +struct tegra_bpmp {
> +	const struct tegra_bpmp_soc *soc;
> +	struct device *dev;
> +
> +	void __iomem *tx_base;
> +	void __iomem *rx_base;
> +
> +	struct {
> +		struct mbox_client client;
> +		struct mbox_chan *channel;
> +	} mbox;
> +
> +	struct tegra_bpmp_channel *channels;
> +	unsigned int num_channels;
> +
> +	struct {
> +		unsigned long *allocated;
> +		unsigned long *busy;
> +		unsigned int count;
> +		struct semaphore lock;
> +	} threaded;
> +
> +	struct list_head mrqs;
> +	spinlock_t lock;
> +
> +	unsigned int num_clocks;
> +	struct clk_hw **clocks;
> +
> +	struct reset_controller_dev rstc;
> +};
> +
> +struct tegra_bpmp *tegra_bpmp_get(struct device *dev);
> +void tegra_bpmp_put(struct tegra_bpmp *bpmp);
> +
> +struct tegra_bpmp_message {
> +	unsigned int mrq;
> +
> +	struct {
> +		const void *data;
> +		size_t size;
> +	} tx;
> +
> +	struct {
> +		void *data;
> +		size_t size;
> +	} rx;
> +};
> +
> +int tegra_bpmp_transfer_atomic(struct tegra_bpmp *bpmp,
> +			       struct tegra_bpmp_message *msg);
> +int tegra_bpmp_transfer(struct tegra_bpmp *bpmp,
> +			struct tegra_bpmp_message *msg);
> +
> +int tegra_bpmp_request_mrq(struct tegra_bpmp *bpmp, unsigned int mrq,
> +			   tegra_bpmp_mrq_handler_t handler, void *data);
> +void tegra_bpmp_free_mrq(struct tegra_bpmp *bpmp, unsigned int mrq,
> +			 void *data);
> +
> +int tegra_bpmp_init_clocks(struct tegra_bpmp *bpmp);
> +int tegra_bpmp_init_resets(struct tegra_bpmp *bpmp);

I believe this will break bisect (because the functions are defined in a
later patch)? But I like the clock/reset driver to refer to bpmp driver
than in the reverse direction.

> +
> +#endif /* __SOC_TEGRA_BPMP_H */
> -- 
> 2.9.0
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [ARM Kernel]     [Linux ARM]     [Linux ARM MSM]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux