linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy
@ 2018-06-05  6:16 Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 1/8] mailbox: ti-msgmgr: Get rid of unused structure members Nishanth Menon
                   ` (7 more replies)
  0 siblings, 8 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Hi,

The following series enables support for Secure Proxy in newest addition in TI's SoC
portfolio - AM654 SoC.

The series is an RFC based off next-20180604 and will post formally once
v4.18-rc1 is available.

The DT and ARCH part of the series is based on https://marc.info/?l=linux-arm-kernel&m=152817866312732&w=2

The series (part 2 of 4) is available here:
https://github.com/nmenon/linux-2.6-playground/commits/upstream/next-20180604/k3-2-am6-sproxy

Full Boot log is available here: https://pastebin.ubuntu.com/p/vWCzMKtBCW/

See AM65x Technical Reference Manual (SPRUID7, April 2018)
for further details: http://www.ti.com/lit/pdf/spruid7

Nishanth Menon (8):
  mailbox: ti-msgmgr: Get rid of unused structure members
  mailbox: ti-msgmgr: Allocate Rx channel resources only on request
  mailbox: ti-msgmgr: Change message count mask to be descriptor based
  mailbox: ti-msgmgr: Move the memory region name to descriptor
  dt-bindings: mailbox: ti,message-manager: Add support for secure proxy
    threads
  mailbox: ti-msgmgr: Add support for Secure Proxy
  drivers: mailbox: Make ti-msgmr driver depend on ARCH_K3
  arm64: dts: ti: k3-am6: Add Secure Proxy instance

 .../bindings/mailbox/ti,message-manager.txt        |  58 +++-
 arch/arm64/boot/dts/ti/k3-am6.dtsi                 |  11 +
 drivers/mailbox/Kconfig                            |   8 +-
 drivers/mailbox/ti-msgmgr.c                        | 353 +++++++++++++++++----
 4 files changed, 351 insertions(+), 79 deletions(-)

-- 
2.15.1

^ permalink raw reply	[flat|nested] 11+ messages in thread

* [RFC PATCH 1/8] mailbox: ti-msgmgr: Get rid of unused structure members
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 2/8] mailbox: ti-msgmgr: Allocate Rx channel resources only on request Nishanth Menon
                   ` (6 subsequent siblings)
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Though q_proxies and q_slices do describe the hardware configuration,
they are not necessary for operation given that the values are
always default. Hence drop the same.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/ti-msgmgr.c | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/drivers/mailbox/ti-msgmgr.c b/drivers/mailbox/ti-msgmgr.c
index 78753a87ba4d..7cd5f9c9c97f 100644
--- a/drivers/mailbox/ti-msgmgr.c
+++ b/drivers/mailbox/ti-msgmgr.c
@@ -42,8 +42,6 @@ struct ti_msgmgr_valid_queue_desc {
  * @queue_count:	Number of Queues
  * @max_message_size:	Message size in bytes
  * @max_messages:	Number of messages
- * @q_slices:		Number of queue engines
- * @q_proxies:		Number of queue proxies per page
  * @data_first_reg:	First data register for proxy data region
  * @data_last_reg:	Last data register for proxy data region
  * @tx_polled:		Do I need to use polled mechanism for tx
@@ -58,8 +56,6 @@ struct ti_msgmgr_desc {
 	u8 queue_count;
 	u8 max_message_size;
 	u8 max_messages;
-	u8 q_slices;
-	u8 q_proxies;
 	u8 data_first_reg;
 	u8 data_last_reg;
 	bool tx_polled;
@@ -494,8 +490,6 @@ static const struct ti_msgmgr_desc k2g_desc = {
 	.queue_count = 64,
 	.max_message_size = 64,
 	.max_messages = 128,
-	.q_slices = 1,
-	.q_proxies = 1,
 	.data_first_reg = 16,
 	.data_last_reg = 31,
 	.tx_polled = false,
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 2/8] mailbox: ti-msgmgr: Allocate Rx channel resources only on request
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 1/8] mailbox: ti-msgmgr: Get rid of unused structure members Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 3/8] mailbox: ti-msgmgr: Change message count mask to be descriptor based Nishanth Menon
                   ` (5 subsequent siblings)
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

In a much bigger system SoCs, the number of Rx channels can be
many and mostly unused based on the system of choice, and not all
Rx channels need IRQs and allocating all memory at probe will be
inefficient. Some SoCs could have total threads in the 100s and usage
would be just 1 Rx thread.

Thus, request and map the IRQs and allocate memory only when needed.

Since these channels are requested by client drivers on need, our
utilization will be optimal.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/ti-msgmgr.c | 91 ++++++++++++++++++++++++++++++---------------
 1 file changed, 61 insertions(+), 30 deletions(-)

diff --git a/drivers/mailbox/ti-msgmgr.c b/drivers/mailbox/ti-msgmgr.c
index 7cd5f9c9c97f..7221590c409c 100644
--- a/drivers/mailbox/ti-msgmgr.c
+++ b/drivers/mailbox/ti-msgmgr.c
@@ -310,6 +310,51 @@ static int ti_msgmgr_send_data(struct mbox_chan *chan, void *data)
 	return 0;
 }
 
+/**
+ *  ti_msgmgr_queue_rx_irq_req() - RX IRQ request
+ *  @dev:	device pointer
+ *  @qinst:	Queue instance
+ *  @chan:	Channel pointer
+ */
+static int ti_msgmgr_queue_rx_irq_req(struct device *dev,
+				      struct ti_queue_inst *qinst,
+				      struct mbox_chan *chan)
+{
+	int ret = 0;
+	char of_rx_irq_name[7];
+	struct device_node *np;
+
+	snprintf(of_rx_irq_name, sizeof(of_rx_irq_name),
+		 "rx_%03d", qinst->queue_id);
+
+	/* Get the IRQ if not found */
+	if (qinst->irq < 0) {
+		np = of_node_get(dev->of_node);
+		if (!np)
+			return -ENODATA;
+		qinst->irq = of_irq_get_byname(np, of_rx_irq_name);
+		of_node_put(np);
+
+		if (qinst->irq < 0) {
+			dev_err(dev,
+				"QID %d PID %d:No IRQ[%s]: %d\n",
+				qinst->queue_id, qinst->proxy_id,
+				of_rx_irq_name, qinst->irq);
+			return qinst->irq;
+		}
+	}
+
+	/* With the expectation that the IRQ might be shared in SoC */
+	ret = request_irq(qinst->irq, ti_msgmgr_queue_rx_interrupt,
+			  IRQF_SHARED, qinst->name, chan);
+	if (ret) {
+		dev_err(dev, "Unable to get IRQ %d on %s(res=%d)\n",
+			qinst->irq, qinst->name, ret);
+	}
+
+	return ret;
+}
+
 /**
  * ti_msgmgr_queue_startup() - Startup queue
  * @chan:	Channel pointer
@@ -318,19 +363,21 @@ static int ti_msgmgr_send_data(struct mbox_chan *chan, void *data)
  */
 static int ti_msgmgr_queue_startup(struct mbox_chan *chan)
 {
-	struct ti_queue_inst *qinst = chan->con_priv;
 	struct device *dev = chan->mbox->dev;
+	struct ti_msgmgr_inst *inst = dev_get_drvdata(dev);
+	struct ti_queue_inst *qinst = chan->con_priv;
+	const struct ti_msgmgr_desc *d = inst->desc;
 	int ret;
 
 	if (!qinst->is_tx) {
-		/*
-		 * With the expectation that the IRQ might be shared in SoC
-		 */
-		ret = request_irq(qinst->irq, ti_msgmgr_queue_rx_interrupt,
-				  IRQF_SHARED, qinst->name, chan);
+		/* Allocate usage buffer for rx */
+		qinst->rx_buff = kzalloc(d->max_message_size, GFP_KERNEL);
+		if (!qinst->rx_buff)
+			return -ENOMEM;
+		/* Request IRQ */
+		ret = ti_msgmgr_queue_rx_irq_req(dev, qinst, chan);
 		if (ret) {
-			dev_err(dev, "Unable to get IRQ %d on %s(res=%d)\n",
-				qinst->irq, qinst->name, ret);
+			kfree(qinst->rx_buff);
 			return ret;
 		}
 	}
@@ -346,8 +393,10 @@ static void ti_msgmgr_queue_shutdown(struct mbox_chan *chan)
 {
 	struct ti_queue_inst *qinst = chan->con_priv;
 
-	if (!qinst->is_tx)
+	if (!qinst->is_tx) {
 		free_irq(qinst->irq, chan);
+		kfree(qinst->rx_buff);
+	}
 }
 
 /**
@@ -425,27 +474,6 @@ static int ti_msgmgr_queue_setup(int idx, struct device *dev,
 		 dev_name(dev), qinst->is_tx ? "tx" : "rx", qinst->queue_id,
 		 qinst->proxy_id);
 
-	if (!qinst->is_tx) {
-		char of_rx_irq_name[7];
-
-		snprintf(of_rx_irq_name, sizeof(of_rx_irq_name),
-			 "rx_%03d", qinst->queue_id);
-
-		qinst->irq = of_irq_get_byname(np, of_rx_irq_name);
-		if (qinst->irq < 0) {
-			dev_crit(dev,
-				 "[%d]QID %d PID %d:No IRQ[%s]: %d\n",
-				 idx, qinst->queue_id, qinst->proxy_id,
-				 of_rx_irq_name, qinst->irq);
-			return qinst->irq;
-		}
-		/* Allocate usage buffer for rx */
-		qinst->rx_buff = devm_kzalloc(dev,
-					      d->max_message_size, GFP_KERNEL);
-		if (!qinst->rx_buff)
-			return -ENOMEM;
-	}
-
 	qinst->queue_buff_start = inst->queue_proxy_region +
 	    Q_DATA_OFFSET(qinst->proxy_id, qinst->queue_id, d->data_first_reg);
 	qinst->queue_buff_end = inst->queue_proxy_region +
@@ -454,6 +482,9 @@ static int ti_msgmgr_queue_setup(int idx, struct device *dev,
 	    Q_STATE_OFFSET(qinst->queue_id);
 	qinst->chan = chan;
 
+	/* Setup an error value for IRQ - Lazy allocation */
+	qinst->irq = -EINVAL;
+
 	chan->con_priv = qinst;
 
 	dev_dbg(dev, "[%d] qidx=%d pidx=%d irq=%d q_s=%p q_e = %p\n",
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 3/8] mailbox: ti-msgmgr: Change message count mask to be descriptor based
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 1/8] mailbox: ti-msgmgr: Get rid of unused structure members Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 2/8] mailbox: ti-msgmgr: Allocate Rx channel resources only on request Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 4/8] mailbox: ti-msgmgr: Move the memory region name to descriptor Nishanth Menon
                   ` (4 subsequent siblings)
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Change mask used to extract the message count to be descriptor based.

This is to support changes for count location for various SoC
solutions.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/ti-msgmgr.c | 25 ++++++++++++++++++-------
 1 file changed, 18 insertions(+), 7 deletions(-)

diff --git a/drivers/mailbox/ti-msgmgr.c b/drivers/mailbox/ti-msgmgr.c
index 7221590c409c..2952339a8446 100644
--- a/drivers/mailbox/ti-msgmgr.c
+++ b/drivers/mailbox/ti-msgmgr.c
@@ -44,6 +44,7 @@ struct ti_msgmgr_valid_queue_desc {
  * @max_messages:	Number of messages
  * @data_first_reg:	First data register for proxy data region
  * @data_last_reg:	Last data register for proxy data region
+ * @status_cnt_mask:	Mask for getting the status value
  * @tx_polled:		Do I need to use polled mechanism for tx
  * @tx_poll_timeout_ms: Timeout in ms if polled
  * @valid_queues:	List of Valid queues that the processor can access
@@ -58,6 +59,7 @@ struct ti_msgmgr_desc {
 	u8 max_messages;
 	u8 data_first_reg;
 	u8 data_last_reg;
+	u32 status_cnt_mask;
 	bool tx_polled;
 	int tx_poll_timeout_ms;
 	const struct ti_msgmgr_valid_queue_desc *valid_queues;
@@ -116,20 +118,24 @@ struct ti_msgmgr_inst {
 
 /**
  * ti_msgmgr_queue_get_num_messages() - Get the number of pending messages
+ * @d:		Description of message manager
  * @qinst:	Queue instance for which we check the number of pending messages
  *
  * Return: number of messages pending in the queue (0 == no pending messages)
  */
-static inline int ti_msgmgr_queue_get_num_messages(struct ti_queue_inst *qinst)
+static inline int
+ti_msgmgr_queue_get_num_messages(const struct ti_msgmgr_desc *d,
+				 struct ti_queue_inst *qinst)
 {
 	u32 val;
+	u32 status_cnt_mask = d->status_cnt_mask;
 
 	/*
 	 * We cannot use relaxed operation here - update may happen
 	 * real-time.
 	 */
-	val = readl(qinst->queue_state) & Q_STATE_ENTRY_COUNT_MASK;
-	val >>= __ffs(Q_STATE_ENTRY_COUNT_MASK);
+	val = readl(qinst->queue_state) & status_cnt_mask;
+	val >>= __ffs(status_cnt_mask);
 
 	return val;
 }
@@ -167,8 +173,9 @@ static irqreturn_t ti_msgmgr_queue_rx_interrupt(int irq, void *p)
 		return IRQ_NONE;
 	}
 
+	desc = inst->desc;
 	/* Do I actually have messages to read? */
-	msg_count = ti_msgmgr_queue_get_num_messages(qinst);
+	msg_count = ti_msgmgr_queue_get_num_messages(desc, qinst);
 	if (!msg_count) {
 		/* Shared IRQ? */
 		dev_dbg(dev, "Spurious event - 0 pending data!\n");
@@ -181,7 +188,6 @@ static irqreturn_t ti_msgmgr_queue_rx_interrupt(int irq, void *p)
 	 * of how many bytes I should be reading. Let the client figure this
 	 * out.. I just read the full message and pass it on..
 	 */
-	desc = inst->desc;
 	message.len = desc->max_message_size;
 	message.buf = (u8 *)qinst->rx_buff;
 
@@ -224,12 +230,14 @@ static irqreturn_t ti_msgmgr_queue_rx_interrupt(int irq, void *p)
 static bool ti_msgmgr_queue_peek_data(struct mbox_chan *chan)
 {
 	struct ti_queue_inst *qinst = chan->con_priv;
+	struct device *dev = chan->mbox->dev;
+	struct ti_msgmgr_inst *inst = dev_get_drvdata(dev);
 	int msg_count;
 
 	if (qinst->is_tx)
 		return false;
 
-	msg_count = ti_msgmgr_queue_get_num_messages(qinst);
+	msg_count = ti_msgmgr_queue_get_num_messages(inst->desc, qinst);
 
 	return msg_count ? true : false;
 }
@@ -243,12 +251,14 @@ static bool ti_msgmgr_queue_peek_data(struct mbox_chan *chan)
 static bool ti_msgmgr_last_tx_done(struct mbox_chan *chan)
 {
 	struct ti_queue_inst *qinst = chan->con_priv;
+	struct device *dev = chan->mbox->dev;
+	struct ti_msgmgr_inst *inst = dev_get_drvdata(dev);
 	int msg_count;
 
 	if (!qinst->is_tx)
 		return false;
 
-	msg_count = ti_msgmgr_queue_get_num_messages(qinst);
+	msg_count = ti_msgmgr_queue_get_num_messages(inst->desc, qinst);
 
 	/* if we have any messages pending.. */
 	return msg_count ? false : true;
@@ -523,6 +533,7 @@ static const struct ti_msgmgr_desc k2g_desc = {
 	.max_messages = 128,
 	.data_first_reg = 16,
 	.data_last_reg = 31,
+	.status_cnt_mask = Q_STATE_ENTRY_COUNT_MASK,
 	.tx_polled = false,
 	.valid_queues = k2g_valid_queues,
 	.num_valid_queues = ARRAY_SIZE(k2g_valid_queues),
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 4/8] mailbox: ti-msgmgr: Move the memory region name to descriptor
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
                   ` (2 preceding siblings ...)
  2018-06-05  6:16 ` [RFC PATCH 3/8] mailbox: ti-msgmgr: Change message count mask to be descriptor based Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads Nishanth Menon
                   ` (3 subsequent siblings)
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

For newer generation of the hardware, the naming of the region is
decided at integration level and there could be additional regions
as well. Hence move the region naming to be described from compatible
descriptor.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/ti-msgmgr.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/mailbox/ti-msgmgr.c b/drivers/mailbox/ti-msgmgr.c
index 2952339a8446..a37d6a4b392f 100644
--- a/drivers/mailbox/ti-msgmgr.c
+++ b/drivers/mailbox/ti-msgmgr.c
@@ -48,6 +48,8 @@ struct ti_msgmgr_valid_queue_desc {
  * @tx_polled:		Do I need to use polled mechanism for tx
  * @tx_poll_timeout_ms: Timeout in ms if polled
  * @valid_queues:	List of Valid queues that the processor can access
+ * @data_region_name:	Name of the proxy data region
+ * @status_region_name:	Name of the proxy status region
  * @num_valid_queues:	Number of valid queues
  *
  * This structure is used in of match data to describe how integration
@@ -63,6 +65,8 @@ struct ti_msgmgr_desc {
 	bool tx_polled;
 	int tx_poll_timeout_ms;
 	const struct ti_msgmgr_valid_queue_desc *valid_queues;
+	const char *data_region_name;
+	const char *status_region_name;
 	int num_valid_queues;
 };
 
@@ -531,6 +535,8 @@ static const struct ti_msgmgr_desc k2g_desc = {
 	.queue_count = 64,
 	.max_message_size = 64,
 	.max_messages = 128,
+	.data_region_name = "queue_proxy_region",
+	.status_region_name = "queue_state_debug_region",
 	.data_first_reg = 16,
 	.data_last_reg = 31,
 	.status_cnt_mask = Q_STATE_ENTRY_COUNT_MASK,
@@ -582,13 +588,13 @@ static int ti_msgmgr_probe(struct platform_device *pdev)
 	inst->desc = desc;
 
 	res = platform_get_resource_byname(pdev, IORESOURCE_MEM,
-					   "queue_proxy_region");
+					   desc->data_region_name);
 	inst->queue_proxy_region = devm_ioremap_resource(dev, res);
 	if (IS_ERR(inst->queue_proxy_region))
 		return PTR_ERR(inst->queue_proxy_region);
 
 	res = platform_get_resource_byname(pdev, IORESOURCE_MEM,
-					   "queue_state_debug_region");
+					   desc->status_region_name);
 	inst->queue_state_debug_region = devm_ioremap_resource(dev, res);
 	if (IS_ERR(inst->queue_state_debug_region))
 		return PTR_ERR(inst->queue_state_debug_region);
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
                   ` (3 preceding siblings ...)
  2018-06-05  6:16 ` [RFC PATCH 4/8] mailbox: ti-msgmgr: Move the memory region name to descriptor Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-12 21:11   ` Rob Herring
  2018-06-05  6:16 ` [RFC PATCH 6/8] mailbox: ti-msgmgr: Add support for Secure Proxy Nishanth Menon
                   ` (2 subsequent siblings)
  7 siblings, 1 reply; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Secure Proxy is another communication scheme in Texas Instrument's
devices intended to provide an unique communication path from various
processors in the System on Chip(SoC) to a central System Controller.

Secure proxy is, in effect, an evolution of current generation Message
Manager hardware block found in K2G devices. However the following
changes have taken place:

Secure Proxy instance exposes "threads" or "proxies" which is
primary representation of "a" communication channel. Each thread is
preconfigured by System controller configuration based on SoC usage
requirements. Secure proxy by itself represents a single "queue" of
communication but allows the proxies to be independently operated.

Each Secure proxy thread can uniquely have their own error and threshold
interrupts allowing for more fine control of IRQ handling.

Provide an hardware description of the same for device tree
representation.

See AM65x Technical Reference Manual (SPRUID7, April 2018)
for further details: http://www.ti.com/lit/pdf/spruid7

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 .../bindings/mailbox/ti,message-manager.txt        | 58 +++++++++++++++++++---
 1 file changed, 50 insertions(+), 8 deletions(-)

diff --git a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
index ebf0e3710cee..de796e90cac6 100644
--- a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
+++ b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
@@ -7,22 +7,40 @@ manager is broken up into queues in different address regions that are called
 "proxies" - each instance is unidirectional and is instantiated at SoC
 integration level to indicate receive or transmit path.
 
+This can also be used to describe Texas Instrument's Secure Proxy
+controller that allows for individually configurable "threads" or
+"proxies" which allow for independent communication scheme.
+
 Message Manager Device Node:
 ===========================
 Required properties:
 --------------------
-- compatible:		Shall be: "ti,k2g-message-manager"
-- reg-names 		queue_proxy_region - Map the queue proxy region.
-			queue_state_debug_region - Map the queue state debug
-			region.
+- compatible:		Shall be one of: "ti,k2g-message-manager",
+			"ti,am654-secure-proxy"
+- reg-names 		for ti,k2g-message-manager, the following shall exist:
+				queue_proxy_region - Map the queue proxy region.
+				queue_state_debug_region - Map the queue state
+					debug region.
+			for ti,am654-secure-proxy, the following shall exist:
+				target_data - Map the proxy data region
+				rt - Map the realtime status region
+				scfg - Map the configuration region
 - reg:			Contains the register map per reg-names.
-- #mbox-cells		Shall be 2. Contains the queue ID and proxy ID in that
-		        order referring to the transfer path.
+- #mbox-cells		for ti,k2g-message-manager, Shall be 2. Contains the
+			queue ID and proxy ID in the following order referring
+			to the transfer path:
+				queue_proxy_region - Map the queue proxy region.
+				queue_state_debug_region - Map the queue state
+					debug region.
+			for ti,am654-secure-proxy, Shall be 1 and shall refer
+		        to the transfer path called thread.
 - interrupt-names:	Contains interrupt names matching the rx transfer path
 			for a given SoC. Receive interrupts shall be of the
-			format: "rx_<QID>".
-			For ti,k2g-message-manager, this shall contain:
+			format:
+			For ti,k2g-message-manager, this shall be: "rx_<QID>"
+			and shall contain:
 				"rx_005", "rx_057"
+			for ti,am654-secure-proxy, this shall be: "rx_<PID>".
 - interrupts:		Contains the interrupt information corresponding to
 			interrupt-names property.
 
@@ -48,3 +66,27 @@ Example(K2G):
 			<&msgmgr 0 0>;
 		[...]
 	};
+
+Example(AM654):
+------------
+
+	secure_proxy: secure_proxy@32c00000 {
+		compatible = "ti,am654-secure-proxy";
+		#mbox-cells = <1>;
+		reg-names = "target_data", "rt", "scfg";
+		reg = <0x0 0x32c00000 0x0 0x100000>,
+		      <0x0 0x32400000 0x0 0x100000>,
+		      <0x0 0x32800000 0x0 0x100000>;
+		interrupt-names = "rx_011";
+		interrupts = <GIC_SPI 32 IRQ_TYPE_LEVEL_HIGH>;
+	};
+
+	dmsc: dmsc {
+		[...]
+		mbox-names = "rx", "tx";
+		# RX Thread ID is 11
+		# TX Thread ID is 13
+		mboxes= <&secure_proxy 11>,
+			<&secure_proxy 13>;
+		[...]
+	};
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 6/8] mailbox: ti-msgmgr: Add support for Secure Proxy
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
                   ` (4 preceding siblings ...)
  2018-06-05  6:16 ` [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 7/8] drivers: mailbox: Make ti-msgmr driver depend on ARCH_K3 Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 8/8] arm64: dts: ti: k3-am6: Add Secure Proxy instance Nishanth Menon
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Secure Proxy is another communication scheme in Texas Instrument's
devices intended to provide an unique communication path from various
processors in the System on Chip(SoC) to a central System Controller.

Secure proxy is, in effect, an evolution of current generation Message
Manager hardware block found in K2G devices. However the following
changes have taken place:

Secure Proxy instance exposes "threads" or "proxies" which is
primary representation of "a" communication channel. Each thread is
preconfigured by System controller configuration based on SoC usage
requirements. Secure proxy by itself represents a single "queue" of
communication but allows the proxies to be independently operated.

Each Secure proxy thread can uniquely have their own error and threshold
interrupts allowing for more fine control of IRQ handling.

Provide the driver support for Secure Proxy and thread instances.

NOTE: Secure proxy configuration is only done by System Controller,
hence these are assumed to be pre-configured instances.

See AM65x Technical Reference Manual (SPRUID7, April 2018)
for further details: http://www.ti.com/lit/pdf/spruid7

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/ti-msgmgr.c | 233 ++++++++++++++++++++++++++++++++++++++------
 1 file changed, 205 insertions(+), 28 deletions(-)

diff --git a/drivers/mailbox/ti-msgmgr.c b/drivers/mailbox/ti-msgmgr.c
index a37d6a4b392f..3e30b80c6401 100644
--- a/drivers/mailbox/ti-msgmgr.c
+++ b/drivers/mailbox/ti-msgmgr.c
@@ -25,6 +25,17 @@
 #define Q_STATE_OFFSET(queue)			((queue) * 0x4)
 #define Q_STATE_ENTRY_COUNT_MASK		(0xFFF000)
 
+#define SPROXY_THREAD_OFFSET(tid) (0x1000 * (tid))
+#define SPROXY_THREAD_DATA_OFFSET(tid, reg) \
+	(SPROXY_THREAD_OFFSET(tid) + ((reg) * 0x4) + 0x4)
+
+#define SPROXY_THREAD_STATUS_OFFSET(tid) (SPROXY_THREAD_OFFSET(tid))
+
+#define SPROXY_THREAD_STATUS_COUNT_MASK (0xFF)
+
+#define SPROXY_THREAD_CTRL_OFFSET(tid) (0x1000 + SPROXY_THREAD_OFFSET(tid))
+#define SPROXY_THREAD_CTRL_DIR_MASK (0x1 << 31)
+
 /**
  * struct ti_msgmgr_valid_queue_desc - SoC valid queues meant for this processor
  * @queue_id:	Queue Number for this path
@@ -45,12 +56,15 @@ struct ti_msgmgr_valid_queue_desc {
  * @data_first_reg:	First data register for proxy data region
  * @data_last_reg:	Last data register for proxy data region
  * @status_cnt_mask:	Mask for getting the status value
+ * @status_err_mask:	Mask for getting the error value, if applicable
  * @tx_polled:		Do I need to use polled mechanism for tx
  * @tx_poll_timeout_ms: Timeout in ms if polled
  * @valid_queues:	List of Valid queues that the processor can access
  * @data_region_name:	Name of the proxy data region
  * @status_region_name:	Name of the proxy status region
+ * @ctrl_region_name:	Name of the proxy control region
  * @num_valid_queues:	Number of valid queues
+ * @is_sproxy:		Is this an Secure Proxy instance?
  *
  * This structure is used in of match data to describe how integration
  * for a specific compatible SoC is done.
@@ -62,12 +76,15 @@ struct ti_msgmgr_desc {
 	u8 data_first_reg;
 	u8 data_last_reg;
 	u32 status_cnt_mask;
+	u32 status_err_mask;
 	bool tx_polled;
 	int tx_poll_timeout_ms;
 	const struct ti_msgmgr_valid_queue_desc *valid_queues;
 	const char *data_region_name;
 	const char *status_region_name;
+	const char *ctrl_region_name;
 	int num_valid_queues;
+	bool is_sproxy;
 };
 
 /**
@@ -80,6 +97,7 @@ struct ti_msgmgr_desc {
  * @queue_buff_start: First register of Data Buffer
  * @queue_buff_end: Last (or confirmation) register of Data buffer
  * @queue_state: Queue status register
+ * @queue_ctrl: Queue Control register
  * @chan:	Mailbox channel
  * @rx_buff:	Receive buffer pointer allocated at probe, max_message_size
  */
@@ -92,6 +110,7 @@ struct ti_queue_inst {
 	void __iomem *queue_buff_start;
 	void __iomem *queue_buff_end;
 	void __iomem *queue_state;
+	void __iomem *queue_ctrl;
 	struct mbox_chan *chan;
 	u32 *rx_buff;
 };
@@ -102,6 +121,7 @@ struct ti_queue_inst {
  * @desc:	Description of the SoC integration
  * @queue_proxy_region:	Queue proxy region where queue buffers are located
  * @queue_state_debug_region:	Queue status register regions
+ * @queue_ctrl_region:	Queue Control register regions
  * @num_valid_queues:	Number of valid queues defined for the processor
  *		Note: other queues are probably reserved for other processors
  *		in the SoC.
@@ -114,6 +134,7 @@ struct ti_msgmgr_inst {
 	const struct ti_msgmgr_desc *desc;
 	void __iomem *queue_proxy_region;
 	void __iomem *queue_state_debug_region;
+	void __iomem *queue_ctrl_region;
 	u8 num_valid_queues;
 	struct ti_queue_inst *qinsts;
 	struct mbox_controller mbox;
@@ -144,6 +165,31 @@ ti_msgmgr_queue_get_num_messages(const struct ti_msgmgr_desc *d,
 	return val;
 }
 
+/**
+ * ti_msgmgr_queue_is_error() - Check to see if there is queue error
+ * @d:		Description of message manager
+ * @qinst:	Queue instance for which we check the number of pending messages
+ *
+ * Return: true if error, else false
+ */
+static inline bool ti_msgmgr_queue_is_error(const struct ti_msgmgr_desc *d,
+					    struct ti_queue_inst *qinst)
+{
+	u32 val;
+
+	/* Msgmgr has no error detection */
+	if (!d->is_sproxy)
+		return false;
+
+	/*
+	 * We cannot use relaxed operation here - update may happen
+	 * real-time.
+	 */
+	val = readl(qinst->queue_state) & d->status_err_mask;
+
+	return val ? true : false;
+}
+
 /**
  * ti_msgmgr_queue_rx_interrupt() - Interrupt handler for receive Queue
  * @irq:	Interrupt number
@@ -178,6 +224,11 @@ static irqreturn_t ti_msgmgr_queue_rx_interrupt(int irq, void *p)
 	}
 
 	desc = inst->desc;
+	if (ti_msgmgr_queue_is_error(desc, qinst)) {
+		dev_err(dev, "Error on Rx channel %s\n", qinst->name);
+		return IRQ_NONE;
+	}
+
 	/* Do I actually have messages to read? */
 	msg_count = ti_msgmgr_queue_get_num_messages(desc, qinst);
 	if (!msg_count) {
@@ -236,12 +287,18 @@ static bool ti_msgmgr_queue_peek_data(struct mbox_chan *chan)
 	struct ti_queue_inst *qinst = chan->con_priv;
 	struct device *dev = chan->mbox->dev;
 	struct ti_msgmgr_inst *inst = dev_get_drvdata(dev);
+	const struct ti_msgmgr_desc *desc = inst->desc;
 	int msg_count;
 
 	if (qinst->is_tx)
 		return false;
 
-	msg_count = ti_msgmgr_queue_get_num_messages(inst->desc, qinst);
+	if (ti_msgmgr_queue_is_error(desc, qinst)) {
+		dev_err(dev, "Error on channel %s\n", qinst->name);
+		return false;
+	}
+
+	msg_count = ti_msgmgr_queue_get_num_messages(desc, qinst);
 
 	return msg_count ? true : false;
 }
@@ -257,12 +314,23 @@ static bool ti_msgmgr_last_tx_done(struct mbox_chan *chan)
 	struct ti_queue_inst *qinst = chan->con_priv;
 	struct device *dev = chan->mbox->dev;
 	struct ti_msgmgr_inst *inst = dev_get_drvdata(dev);
+	const struct ti_msgmgr_desc *desc = inst->desc;
 	int msg_count;
 
 	if (!qinst->is_tx)
 		return false;
 
-	msg_count = ti_msgmgr_queue_get_num_messages(inst->desc, qinst);
+	if (ti_msgmgr_queue_is_error(desc, qinst)) {
+		dev_err(dev, "Error on channel %s\n", qinst->name);
+		return false;
+	}
+
+	msg_count = ti_msgmgr_queue_get_num_messages(desc, qinst);
+
+	if (desc->is_sproxy) {
+		/* In secure proxy, msg_count indicates how many we can send */
+		return msg_count ? true : false;
+	}
 
 	/* if we have any messages pending.. */
 	return msg_count ? false : true;
@@ -292,6 +360,11 @@ static int ti_msgmgr_send_data(struct mbox_chan *chan, void *data)
 	}
 	desc = inst->desc;
 
+	if (ti_msgmgr_queue_is_error(desc, qinst)) {
+		dev_err(dev, "Error on channel %s\n", qinst->name);
+		return false;
+	}
+
 	if (desc->max_message_size < message->len) {
 		dev_err(dev, "Queue %s message length %zu > max %d\n",
 			qinst->name, message->len, desc->max_message_size);
@@ -327,10 +400,12 @@ static int ti_msgmgr_send_data(struct mbox_chan *chan, void *data)
 /**
  *  ti_msgmgr_queue_rx_irq_req() - RX IRQ request
  *  @dev:	device pointer
+ *  @d:		descriptor for ti_msgmgr
  *  @qinst:	Queue instance
  *  @chan:	Channel pointer
  */
 static int ti_msgmgr_queue_rx_irq_req(struct device *dev,
+				      const struct ti_msgmgr_desc *d,
 				      struct ti_queue_inst *qinst,
 				      struct mbox_chan *chan)
 {
@@ -339,7 +414,7 @@ static int ti_msgmgr_queue_rx_irq_req(struct device *dev,
 	struct device_node *np;
 
 	snprintf(of_rx_irq_name, sizeof(of_rx_irq_name),
-		 "rx_%03d", qinst->queue_id);
+		 "rx_%03d", d->is_sproxy ? qinst->proxy_id : qinst->queue_id);
 
 	/* Get the IRQ if not found */
 	if (qinst->irq < 0) {
@@ -382,6 +457,24 @@ static int ti_msgmgr_queue_startup(struct mbox_chan *chan)
 	struct ti_queue_inst *qinst = chan->con_priv;
 	const struct ti_msgmgr_desc *d = inst->desc;
 	int ret;
+	int msg_count;
+
+	/*
+	 * If sproxy is starting and can send messages, we are a Tx thread,
+	 * else Rx
+	 */
+	if (d->is_sproxy) {
+		qinst->is_tx = (readl(qinst->queue_ctrl) &
+				SPROXY_THREAD_CTRL_DIR_MASK) ? false : true;
+
+		msg_count = ti_msgmgr_queue_get_num_messages(d, qinst);
+
+		if (!msg_count && qinst->is_tx) {
+			dev_err(dev, "%s: Cannot transmit with 0 credits!\n",
+				qinst->name);
+			return -EINVAL;
+		}
+	}
 
 	if (!qinst->is_tx) {
 		/* Allocate usage buffer for rx */
@@ -389,7 +482,7 @@ static int ti_msgmgr_queue_startup(struct mbox_chan *chan)
 		if (!qinst->rx_buff)
 			return -ENOMEM;
 		/* Request IRQ */
-		ret = ti_msgmgr_queue_rx_irq_req(dev, qinst, chan);
+		ret = ti_msgmgr_queue_rx_irq_req(dev, d, qinst, chan);
 		if (ret) {
 			kfree(qinst->rx_buff);
 			return ret;
@@ -427,20 +520,38 @@ static struct mbox_chan *ti_msgmgr_of_xlate(struct mbox_controller *mbox,
 	struct ti_msgmgr_inst *inst;
 	int req_qid, req_pid;
 	struct ti_queue_inst *qinst;
-	int i;
+	const struct ti_msgmgr_desc *d;
+	int i, ncells;
 
 	inst = container_of(mbox, struct ti_msgmgr_inst, mbox);
 	if (WARN_ON(!inst))
 		return ERR_PTR(-EINVAL);
 
-	/* #mbox-cells is 2 */
-	if (p->args_count != 2) {
-		dev_err(inst->dev, "Invalid arguments in dt[%d] instead of 2\n",
-			p->args_count);
+	d = inst->desc;
+
+	if (d->is_sproxy)
+		ncells = 1;
+	else
+		ncells = 2;
+	if (p->args_count != ncells) {
+		dev_err(inst->dev, "Invalid arguments in dt[%d]. Must be %d\n",
+			p->args_count, ncells);
 		return ERR_PTR(-EINVAL);
 	}
-	req_qid = p->args[0];
-	req_pid = p->args[1];
+	if (ncells == 1) {
+		req_qid = 0;
+		req_pid = p->args[0];
+	} else {
+		req_qid = p->args[0];
+		req_pid = p->args[1];
+	}
+
+	if (d->is_sproxy) {
+		if (req_pid > d->num_valid_queues)
+			goto err;
+		qinst = &inst->qinsts[req_pid];
+		return qinst->chan;
+	}
 
 	for (qinst = inst->qinsts, i = 0; i < inst->num_valid_queues;
 	     i++, qinst++) {
@@ -448,6 +559,7 @@ static struct mbox_chan *ti_msgmgr_of_xlate(struct mbox_controller *mbox,
 			return qinst->chan;
 	}
 
+err:
 	dev_err(inst->dev, "Queue ID %d, Proxy ID %d is wrong on %s\n",
 		req_qid, req_pid, p->np->name);
 	return ERR_PTR(-ENOENT);
@@ -474,6 +586,8 @@ static int ti_msgmgr_queue_setup(int idx, struct device *dev,
 				 struct ti_queue_inst *qinst,
 				 struct mbox_chan *chan)
 {
+	char *dir;
+
 	qinst->proxy_id = qd->proxy_id;
 	qinst->queue_id = qd->queue_id;
 
@@ -483,17 +597,38 @@ static int ti_msgmgr_queue_setup(int idx, struct device *dev,
 		return -ERANGE;
 	}
 
-	qinst->is_tx = qd->is_tx;
-	snprintf(qinst->name, sizeof(qinst->name), "%s %s_%03d_%03d",
-		 dev_name(dev), qinst->is_tx ? "tx" : "rx", qinst->queue_id,
-		 qinst->proxy_id);
-
-	qinst->queue_buff_start = inst->queue_proxy_region +
-	    Q_DATA_OFFSET(qinst->proxy_id, qinst->queue_id, d->data_first_reg);
-	qinst->queue_buff_end = inst->queue_proxy_region +
-	    Q_DATA_OFFSET(qinst->proxy_id, qinst->queue_id, d->data_last_reg);
-	qinst->queue_state = inst->queue_state_debug_region +
-	    Q_STATE_OFFSET(qinst->queue_id);
+	if (d->is_sproxy) {
+		qinst->queue_buff_start = inst->queue_proxy_region +
+		    SPROXY_THREAD_DATA_OFFSET(qinst->proxy_id,
+					      d->data_first_reg);
+		qinst->queue_buff_end = inst->queue_proxy_region +
+		    SPROXY_THREAD_DATA_OFFSET(qinst->proxy_id,
+					      d->data_last_reg);
+		qinst->queue_state = inst->queue_state_debug_region +
+		    SPROXY_THREAD_STATUS_OFFSET(qinst->proxy_id);
+		qinst->queue_ctrl = inst->queue_ctrl_region +
+		    SPROXY_THREAD_CTRL_OFFSET(qinst->proxy_id);
+
+		/* XXX: DONOT read registers here!.. Some may be unusable */
+		dir = "thr";
+		snprintf(qinst->name, sizeof(qinst->name), "%s %s_%03d",
+			 dev_name(dev), dir, qinst->proxy_id);
+	} else {
+		qinst->queue_buff_start = inst->queue_proxy_region +
+		    Q_DATA_OFFSET(qinst->proxy_id, qinst->queue_id,
+				  d->data_first_reg);
+		qinst->queue_buff_end = inst->queue_proxy_region +
+		    Q_DATA_OFFSET(qinst->proxy_id, qinst->queue_id,
+				  d->data_last_reg);
+		qinst->queue_state =
+		    inst->queue_state_debug_region +
+		    Q_STATE_OFFSET(qinst->queue_id);
+		qinst->is_tx = qd->is_tx;
+		dir = qinst->is_tx ? "tx" : "rx";
+		snprintf(qinst->name, sizeof(qinst->name), "%s %s_%03d_%03d",
+			 dev_name(dev), dir, qinst->queue_id, qinst->proxy_id);
+	}
+
 	qinst->chan = chan;
 
 	/* Setup an error value for IRQ - Lazy allocation */
@@ -543,12 +678,29 @@ static const struct ti_msgmgr_desc k2g_desc = {
 	.tx_polled = false,
 	.valid_queues = k2g_valid_queues,
 	.num_valid_queues = ARRAY_SIZE(k2g_valid_queues),
+	.is_sproxy = false,
+};
+
+static const struct ti_msgmgr_desc am654_desc = {
+	.queue_count = 190,
+	.num_valid_queues = 190,
+	.max_message_size = 60,
+	.data_region_name = "target_data",
+	.status_region_name = "rt",
+	.ctrl_region_name = "scfg",
+	.data_first_reg = 0,
+	.data_last_reg = 14,
+	.status_cnt_mask = SPROXY_THREAD_STATUS_COUNT_MASK,
+	.tx_polled = false,
+	.is_sproxy = true,
 };
 
 static const struct of_device_id ti_msgmgr_of_match[] = {
 	{.compatible = "ti,k2g-message-manager", .data = &k2g_desc},
+	{.compatible = "ti,am654-secure-proxy", .data = &am654_desc},
 	{ /* Sentinel */ }
 };
+
 MODULE_DEVICE_TABLE(of, ti_msgmgr_of_match);
 
 static int ti_msgmgr_probe(struct platform_device *pdev)
@@ -599,6 +751,14 @@ static int ti_msgmgr_probe(struct platform_device *pdev)
 	if (IS_ERR(inst->queue_state_debug_region))
 		return PTR_ERR(inst->queue_state_debug_region);
 
+	if (desc->is_sproxy) {
+		res = platform_get_resource_byname(pdev, IORESOURCE_MEM,
+						   desc->ctrl_region_name);
+		inst->queue_ctrl_region = devm_ioremap_resource(dev, res);
+		if (IS_ERR(inst->queue_ctrl_region))
+			return PTR_ERR(inst->queue_ctrl_region);
+	}
+
 	dev_dbg(dev, "proxy region=%p, queue_state=%p\n",
 		inst->queue_proxy_region, inst->queue_state_debug_region);
 
@@ -620,12 +780,29 @@ static int ti_msgmgr_probe(struct platform_device *pdev)
 		return -ENOMEM;
 	inst->chans = chans;
 
-	for (i = 0, queue_desc = desc->valid_queues;
-	     i < queue_count; i++, qinst++, chans++, queue_desc++) {
-		ret = ti_msgmgr_queue_setup(i, dev, np, inst,
-					    desc, queue_desc, qinst, chans);
-		if (ret)
-			return ret;
+	if (desc->is_sproxy) {
+		struct ti_msgmgr_valid_queue_desc sproxy_desc;
+
+		/* All proxies may be valid in Secure Proxy instance */
+		for (i = 0; i < queue_count; i++, qinst++, chans++) {
+			sproxy_desc.queue_id = 0;
+			sproxy_desc.proxy_id = i;
+			ret = ti_msgmgr_queue_setup(i, dev, np, inst,
+						    desc, &sproxy_desc, qinst,
+						    chans);
+			if (ret)
+				return ret;
+		}
+	} else {
+		/* Only Some proxies are valid in Message Manager */
+		for (i = 0, queue_desc = desc->valid_queues;
+		     i < queue_count; i++, qinst++, chans++, queue_desc++) {
+			ret = ti_msgmgr_queue_setup(i, dev, np, inst,
+						    desc, queue_desc, qinst,
+						    chans);
+			if (ret)
+				return ret;
+		}
 	}
 
 	mbox = &inst->mbox;
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 7/8] drivers: mailbox: Make ti-msgmr driver depend on ARCH_K3
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
                   ` (5 preceding siblings ...)
  2018-06-05  6:16 ` [RFC PATCH 6/8] mailbox: ti-msgmgr: Add support for Secure Proxy Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  2018-06-05  6:16 ` [RFC PATCH 8/8] arm64: dts: ti: k3-am6: Add Secure Proxy instance Nishanth Menon
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

ti-msgmr driver can support K3 platforms as well.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 drivers/mailbox/Kconfig | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
index 725dce5ba62d..f87a857d21a5 100644
--- a/drivers/mailbox/Kconfig
+++ b/drivers/mailbox/Kconfig
@@ -99,12 +99,12 @@ config STI_MBOX
 
 config TI_MESSAGE_MANAGER
 	tristate "Texas Instruments Message Manager Driver"
-	depends on ARCH_KEYSTONE
+	depends on ARCH_KEYSTONE || ARCH_K3
 	help
 	  An implementation of Message Manager slave driver for Keystone
-	  architecture SoCs from Texas Instruments. Message Manager is a
-	  communication entity found on few of Texas Instrument's keystone
-	  architecture SoCs. These may be used for communication between
+	  and K3 architecture SoCs from Texas Instruments. Message Manager
+	  is a communication entity found on few of Texas Instrument's keystone
+	  and K3 architecture SoCs. These may be used for communication between
 	  multiple processors within the SoC. Select this driver if your
 	  platform has support for the hardware block.
 
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* [RFC PATCH 8/8] arm64: dts: ti: k3-am6: Add Secure Proxy instance
  2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
                   ` (6 preceding siblings ...)
  2018-06-05  6:16 ` [RFC PATCH 7/8] drivers: mailbox: Make ti-msgmr driver depend on ARCH_K3 Nishanth Menon
@ 2018-06-05  6:16 ` Nishanth Menon
  7 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-05  6:16 UTC (permalink / raw)
  To: Jassi Brar, Rob Herring
  Cc: linux-arm-kernel, linux-kernel, devicetree, Nishanth Menon,
	Tero Kristo, Suman Anna

Add support for Secure Proxy instances in AM6 family of SoCs.

Signed-off-by: Nishanth Menon <nm@ti.com>
---
 arch/arm64/boot/dts/ti/k3-am6.dtsi | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/arch/arm64/boot/dts/ti/k3-am6.dtsi b/arch/arm64/boot/dts/ti/k3-am6.dtsi
index cdfa12173aac..07e9cc05519c 100644
--- a/arch/arm64/boot/dts/ti/k3-am6.dtsi
+++ b/arch/arm64/boot/dts/ti/k3-am6.dtsi
@@ -86,6 +86,17 @@
 			};
 		};
 
+		secure_proxy: secure_proxy@32c00000 {
+			compatible = "ti,am654-secure-proxy";
+			#mbox-cells = <1>;
+			reg-names = "target_data", "rt", "scfg";
+			reg = <0x0 0x32c00000 0x0 0x100000>,
+			      <0x0 0x32400000 0x0 0x100000>,
+			      <0x0 0x32800000 0x0 0x100000>;
+			interrupt-names = "rx_011";
+			interrupts = <GIC_SPI 37 IRQ_TYPE_LEVEL_HIGH>;
+		};
+
 		wkup_uart0: serial@42300000 {
 			compatible = "ti,am654-uart", "ti,omap4-uart", "ns16550a";
 			reg = <0x0 0x42300000 0x0 0x100>;
-- 
2.15.1

^ permalink raw reply related	[flat|nested] 11+ messages in thread

* Re: [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads
  2018-06-05  6:16 ` [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads Nishanth Menon
@ 2018-06-12 21:11   ` Rob Herring
  2018-06-12 22:04     ` Nishanth Menon
  0 siblings, 1 reply; 11+ messages in thread
From: Rob Herring @ 2018-06-12 21:11 UTC (permalink / raw)
  To: Nishanth Menon
  Cc: Jassi Brar, linux-arm-kernel, linux-kernel, devicetree,
	Tero Kristo, Suman Anna

On Tue, Jun 05, 2018 at 01:16:26AM -0500, Nishanth Menon wrote:
> Secure Proxy is another communication scheme in Texas Instrument's
> devices intended to provide an unique communication path from various
> processors in the System on Chip(SoC) to a central System Controller.
> 
> Secure proxy is, in effect, an evolution of current generation Message
> Manager hardware block found in K2G devices. However the following
> changes have taken place:
> 
> Secure Proxy instance exposes "threads" or "proxies" which is
> primary representation of "a" communication channel. Each thread is
> preconfigured by System controller configuration based on SoC usage
> requirements. Secure proxy by itself represents a single "queue" of
> communication but allows the proxies to be independently operated.
> 
> Each Secure proxy thread can uniquely have their own error and threshold
> interrupts allowing for more fine control of IRQ handling.
> 
> Provide an hardware description of the same for device tree
> representation.
> 
> See AM65x Technical Reference Manual (SPRUID7, April 2018)
> for further details: http://www.ti.com/lit/pdf/spruid7
> 
> Signed-off-by: Nishanth Menon <nm@ti.com>
> ---
>  .../bindings/mailbox/ti,message-manager.txt        | 58 +++++++++++++++++++---
>  1 file changed, 50 insertions(+), 8 deletions(-)
> 
> diff --git a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> index ebf0e3710cee..de796e90cac6 100644
> --- a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> +++ b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> @@ -7,22 +7,40 @@ manager is broken up into queues in different address regions that are called
>  "proxies" - each instance is unidirectional and is instantiated at SoC
>  integration level to indicate receive or transmit path.
>  
> +This can also be used to describe Texas Instrument's Secure Proxy
> +controller that allows for individually configurable "threads" or
> +"proxies" which allow for independent communication scheme.

There seems to be very little re-use here. I think I would make this 
its own doc.

Rob

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads
  2018-06-12 21:11   ` Rob Herring
@ 2018-06-12 22:04     ` Nishanth Menon
  0 siblings, 0 replies; 11+ messages in thread
From: Nishanth Menon @ 2018-06-12 22:04 UTC (permalink / raw)
  To: Rob Herring
  Cc: Jassi Brar, linux-arm-kernel, linux-kernel, devicetree,
	Tero Kristo, Suman Anna

On 21:11-20180612, Rob Herring wrote:
[...]
> > diff --git a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> > index ebf0e3710cee..de796e90cac6 100644
> > --- a/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> > +++ b/Documentation/devicetree/bindings/mailbox/ti,message-manager.txt
> > @@ -7,22 +7,40 @@ manager is broken up into queues in different address regions that are called
> >  "proxies" - each instance is unidirectional and is instantiated at SoC
> >  integration level to indicate receive or transmit path.
> >  
> > +This can also be used to describe Texas Instrument's Secure Proxy
> > +controller that allows for individually configurable "threads" or
> > +"proxies" which allow for independent communication scheme.
> 
> There seems to be very little re-use here. I think I would make this 
> its own doc.

Agreed. will do the same in the formal series. Thanks.
-- 
Regards,
Nishanth Menon

^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2018-06-12 22:04 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-05  6:16 [RFC PATCH 0/8] mailbox: ti-msgmgr: Add support for AM654 Secure Proxy Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 1/8] mailbox: ti-msgmgr: Get rid of unused structure members Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 2/8] mailbox: ti-msgmgr: Allocate Rx channel resources only on request Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 3/8] mailbox: ti-msgmgr: Change message count mask to be descriptor based Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 4/8] mailbox: ti-msgmgr: Move the memory region name to descriptor Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 5/8] dt-bindings: mailbox: ti,message-manager: Add support for secure proxy threads Nishanth Menon
2018-06-12 21:11   ` Rob Herring
2018-06-12 22:04     ` Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 6/8] mailbox: ti-msgmgr: Add support for Secure Proxy Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 7/8] drivers: mailbox: Make ti-msgmr driver depend on ARCH_K3 Nishanth Menon
2018-06-05  6:16 ` [RFC PATCH 8/8] arm64: dts: ti: k3-am6: Add Secure Proxy instance Nishanth Menon

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).