[PATCH v5 0/5] drivers: remoteproc: xilinx: add mailbox support

[PATCH v5 0/5] drivers: remoteproc: xilinx: add mailbox support

[Tanmay Shah]

This series adds mailbox support required for rpmsg communication
in Xilinx remoteproc driver. It also fixes few bugs in Xilinx
zynqmp-ipi driver that handles corner cases.

---

Changes in v5:
  - mailbox node count should not be 0 for driver to probe
  - Remove spurious change
  - Include kick op regardless status of mailbox nodes in dts

Changes in v4:
  - split mailbox driver patch into 3 different patches
  - use of_phandle_iterator_next to get memory-region nodes
  - move zynqmp_r5_setup_mbox from zynqmp_r5_add_rproc_core to
    zynqmp_r5_cluster_init

Changes in v3:
  - split single patch into multiple patches
  - add appropriate fixes tag for patches fixing previous bug in the driver
  - make enhancements in zynqmp-ipi driver that handles corener cases
  - fix memory-region carveout names
  - fix multi-line comment format
  - do not mixup mailbox information with memory-regions
  - fix redundant dev_warn for split mode
  - Setting up mailboxes should return an error code
  - Move mailbox setup during driver probe
  - Add .kick function only if mailbox setup is success

v2: https://lore.kernel.org/all/20230126213154.1707300-1-tanmay.shah@amd.com/


Tanmay Shah (5):
  mailbox: zynqmp: fix counts of child nodes
  mailbox: zynqmp: fix IPI isr handling
  mailbox: zynqmp: fix typo in IPI documentation
  drivers: remoteproc: xilinx: fix carveout names
  remoteproc: xilinx: add mailbox channels for rpmsg

 drivers/mailbox/zynqmp-ipi-mailbox.c       |  13 +-
 drivers/remoteproc/xlnx_r5_remoteproc.c    | 308 ++++++++++++++++-----
 include/linux/mailbox/zynqmp-ipi-message.h |   2 +-
 3 files changed, 247 insertions(+), 76 deletions(-)


base-commit: fe15c26ee26efa11741a7b632e9f23b01aca4cc6
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH v5 1/5] mailbox: zynqmp: fix counts of child nodes

[Tanmay Shah]

If child mailbox node status is disabled it causes
crash in interrupt handler. Fix this by assigning
only available child node during driver probe.

Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---

Changes in v5:
  - mailbox node count should not be 0 for driver to probe

 drivers/mailbox/zynqmp-ipi-mailbox.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/mailbox/zynqmp-ipi-mailbox.c b/drivers/mailbox/zynqmp-ipi-mailbox.c
index a4c8d23c76e2..ae47fbac38d2 100644
--- a/drivers/mailbox/zynqmp-ipi-mailbox.c
+++ b/drivers/mailbox/zynqmp-ipi-mailbox.c
@@ -634,7 +634,12 @@ static int zynqmp_ipi_probe(struct platform_device *pdev)
 	struct zynqmp_ipi_mbox *mbox;
 	int num_mboxes, ret = -EINVAL;
 
-	num_mboxes = of_get_child_count(np);
+	num_mboxes = of_get_available_child_count(np);
+	if (num_mboxes == 0) {
+		dev_err(dev, "mailbox nodes not available\n");
+		return -EINVAL;
+	}
+
 	pdata = devm_kzalloc(dev, struct_size(pdata, ipi_mboxes, num_mboxes),
 			     GFP_KERNEL);
 	if (!pdata)
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH v5 1/5] mailbox: zynqmp: fix counts of child nodes

[Michal Simek]

On 3/11/23 02:24, Tanmay Shah wrote:
> If child mailbox node status is disabled it causes
> crash in interrupt handler. Fix this by assigning
> only available child node during driver probe.
> 
> Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> ---
> 
> Changes in v5:
>    - mailbox node count should not be 0 for driver to probe
> 
>   drivers/mailbox/zynqmp-ipi-mailbox.c | 7 ++++++-
>   1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mailbox/zynqmp-ipi-mailbox.c b/drivers/mailbox/zynqmp-ipi-mailbox.c
> index a4c8d23c76e2..ae47fbac38d2 100644
> --- a/drivers/mailbox/zynqmp-ipi-mailbox.c
> +++ b/drivers/mailbox/zynqmp-ipi-mailbox.c
> @@ -634,7 +634,12 @@ static int zynqmp_ipi_probe(struct platform_device *pdev)
>   	struct zynqmp_ipi_mbox *mbox;
>   	int num_mboxes, ret = -EINVAL;
>   
> -	num_mboxes = of_get_child_count(np);
> +	num_mboxes = of_get_available_child_count(np);
> +	if (num_mboxes == 0) {
> +		dev_err(dev, "mailbox nodes not available\n");
> +		return -EINVAL;
> +	}
> +
>   	pdata = devm_kzalloc(dev, struct_size(pdata, ipi_mboxes, num_mboxes),
>   			     GFP_KERNEL);
>   	if (!pdata)

Acked-by: Michal Simek <michal.simek@amd.com>

Thanks,
Michal

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH v5 1/5] mailbox: zynqmp: fix counts of child nodes

[Mathieu Poirier]

On Fri, Mar 10, 2023 at 05:24:04PM -0800, Tanmay Shah wrote:
> If child mailbox node status is disabled it causes
> crash in interrupt handler. Fix this by assigning
> only available child node during driver probe.
> 
> Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> ---
> 
> Changes in v5:
>   - mailbox node count should not be 0 for driver to probe
> 
>  drivers/mailbox/zynqmp-ipi-mailbox.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mailbox/zynqmp-ipi-mailbox.c b/drivers/mailbox/zynqmp-ipi-mailbox.c
> index a4c8d23c76e2..ae47fbac38d2 100644
> --- a/drivers/mailbox/zynqmp-ipi-mailbox.c
> +++ b/drivers/mailbox/zynqmp-ipi-mailbox.c
> @@ -634,7 +634,12 @@ static int zynqmp_ipi_probe(struct platform_device *pdev)
>  	struct zynqmp_ipi_mbox *mbox;
>  	int num_mboxes, ret = -EINVAL;
>  
> -	num_mboxes = of_get_child_count(np);
> +	num_mboxes = of_get_available_child_count(np);
> +	if (num_mboxes == 0) {
> +		dev_err(dev, "mailbox nodes not available\n");
> +		return -EINVAL;
> +	}
> +

This patchset looks good now.  I'll queue it once I get an ack from Michal.

Thanks,
Mathieu

>  	pdata = devm_kzalloc(dev, struct_size(pdata, ipi_mboxes, num_mboxes),
>  			     GFP_KERNEL);
>  	if (!pdata)
> -- 
> 2.25.1
> 

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH v5 1/5] mailbox: zynqmp: fix counts of child nodes

[Michal Simek]

On 3/13/23 20:38, Mathieu Poirier wrote:
> On Fri, Mar 10, 2023 at 05:24:04PM -0800, Tanmay Shah wrote:
>> If child mailbox node status is disabled it causes
>> crash in interrupt handler. Fix this by assigning
>> only available child node during driver probe.
>>
>> Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
>> ---
>>
>> Changes in v5:
>>    - mailbox node count should not be 0 for driver to probe
>>
>>   drivers/mailbox/zynqmp-ipi-mailbox.c | 7 ++++++-
>>   1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/mailbox/zynqmp-ipi-mailbox.c b/drivers/mailbox/zynqmp-ipi-mailbox.c
>> index a4c8d23c76e2..ae47fbac38d2 100644
>> --- a/drivers/mailbox/zynqmp-ipi-mailbox.c
>> +++ b/drivers/mailbox/zynqmp-ipi-mailbox.c
>> @@ -634,7 +634,12 @@ static int zynqmp_ipi_probe(struct platform_device *pdev)
>>   	struct zynqmp_ipi_mbox *mbox;
>>   	int num_mboxes, ret = -EINVAL;
>>   
>> -	num_mboxes = of_get_child_count(np);
>> +	num_mboxes = of_get_available_child_count(np);
>> +	if (num_mboxes == 0) {
>> +		dev_err(dev, "mailbox nodes not available\n");
>> +		return -EINVAL;
>> +	}
>> +
> 
> This patchset looks good now.  I'll queue it once I get an ack from Michal.

Patches 1,2,3 are acked by me already. 4,5 are remoteproc one which I am not 
testing.
If 4,5 are fine please queue them.

Thanks,
Michal

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH v5 2/5] mailbox: zynqmp: fix IPI isr handling

[Tanmay Shah]

Multiple IPI channels are mapped to same interrupt handler.
Current isr implementation handles only one channel per isr.
Fix this behavior by checking isr status bit of all child
mailbox nodes.

Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
Acked-by: Michal Simek <michal.simek@amd.com>
---
 drivers/mailbox/zynqmp-ipi-mailbox.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/mailbox/zynqmp-ipi-mailbox.c b/drivers/mailbox/zynqmp-ipi-mailbox.c
index ae47fbac38d2..d097f45b0e5f 100644
--- a/drivers/mailbox/zynqmp-ipi-mailbox.c
+++ b/drivers/mailbox/zynqmp-ipi-mailbox.c
@@ -152,7 +152,7 @@ static irqreturn_t zynqmp_ipi_interrupt(int irq, void *data)
 	struct zynqmp_ipi_message *msg;
 	u64 arg0, arg3;
 	struct arm_smccc_res res;
-	int ret, i;
+	int ret, i, status = IRQ_NONE;
 
 	(void)irq;
 	arg0 = SMC_IPI_MAILBOX_STATUS_ENQUIRY;
@@ -170,11 +170,11 @@ static irqreturn_t zynqmp_ipi_interrupt(int irq, void *data)
 				memcpy_fromio(msg->data, mchan->req_buf,
 					      msg->len);
 				mbox_chan_received_data(chan, (void *)msg);
-				return IRQ_HANDLED;
+				status = IRQ_HANDLED;
 			}
 		}
 	}
-	return IRQ_NONE;
+	return status;
 }
 
 /**
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH v5 3/5] mailbox: zynqmp: fix typo in IPI documentation

[Tanmay Shah]

Xilinx IPI message buffers allows 32-byte data transfer.
Fix documentation that says 12 bytes

Fixes: 4981b82ba2ff ("mailbox: ZynqMP IPI mailbox controller")
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
Acked-by: Michal Simek <michal.simek@amd.com>
---
 include/linux/mailbox/zynqmp-ipi-message.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/mailbox/zynqmp-ipi-message.h b/include/linux/mailbox/zynqmp-ipi-message.h
index 35ce84c8ca02..31d8046d945e 100644
--- a/include/linux/mailbox/zynqmp-ipi-message.h
+++ b/include/linux/mailbox/zynqmp-ipi-message.h
@@ -9,7 +9,7 @@
  * @data: message payload
  *
  * This is the structure for data used in mbox_send_message
- * the maximum length of data buffer is fixed to 12 bytes.
+ * the maximum length of data buffer is fixed to 32 bytes.
  * Client is supposed to be aware of this.
  */
 struct zynqmp_ipi_message {
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH v5 4/5] drivers: remoteproc: xilinx: fix carveout names

[Tanmay Shah]

If the unit address is appended to node name of memory-region,
then adding rproc carveouts fails as node name and unit-address
both are passed as carveout name (i.e. vdev0vring0@xxxxxxxx). However,
only node name is expected by remoteproc framework. This patch moves
memory-region node parsing from driver probe to prepare and
only passes node-name and not unit-address

Fixes: 6b291e8020a8 ("drivers: remoteproc: Add Xilinx r5 remoteproc driver")
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---
 drivers/remoteproc/xlnx_r5_remoteproc.c | 90 ++++++-------------------
 1 file changed, 20 insertions(+), 70 deletions(-)

diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
index 2db57d394155..5dd007622603 100644
--- a/drivers/remoteproc/xlnx_r5_remoteproc.c
+++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
@@ -61,8 +61,6 @@ static const struct mem_bank_data zynqmp_tcm_banks[] = {
  * @np: device node of RPU instance
  * @tcm_bank_count: number TCM banks accessible to this RPU
  * @tcm_banks: array of each TCM bank data
- * @rmem_count: Number of reserved mem regions
- * @rmem: reserved memory region nodes from device tree
  * @rproc: rproc handle
  * @pm_domain_id: RPU CPU power domain id
  */
@@ -71,8 +69,6 @@ struct zynqmp_r5_core {
 	struct device_node *np;
 	int tcm_bank_count;
 	struct mem_bank_data **tcm_banks;
-	int rmem_count;
-	struct reserved_mem **rmem;
 	struct rproc *rproc;
 	u32 pm_domain_id;
 };
@@ -239,21 +235,29 @@ static int add_mem_regions_carveout(struct rproc *rproc)
 {
 	struct rproc_mem_entry *rproc_mem;
 	struct zynqmp_r5_core *r5_core;
+	struct of_phandle_iterator it;
 	struct reserved_mem *rmem;
-	int i, num_mem_regions;
+	int i = 0;
 
 	r5_core = (struct zynqmp_r5_core *)rproc->priv;
-	num_mem_regions = r5_core->rmem_count;
 
-	for (i = 0; i < num_mem_regions; i++) {
-		rmem = r5_core->rmem[i];
+	/* Register associated reserved memory regions */
+	of_phandle_iterator_init(&it, r5_core->np, "memory-region", NULL, 0);
 
-		if (!strncmp(rmem->name, "vdev0buffer", strlen("vdev0buffer"))) {
+	while (of_phandle_iterator_next(&it) == 0) {
+		rmem = of_reserved_mem_lookup(it.node);
+		if (!rmem) {
+			of_node_put(it.node);
+			dev_err(&rproc->dev, "unable to acquire memory-region\n");
+			return -EINVAL;
+		}
+
+		if (!strcmp(it.node->name, "vdev0buffer")) {
 			/* Init reserved memory for vdev buffer */
 			rproc_mem = rproc_of_resm_mem_entry_init(&rproc->dev, i,
 								 rmem->size,
 								 rmem->base,
-								 rmem->name);
+								 it.node->name);
 		} else {
 			/* Register associated reserved memory regions */
 			rproc_mem = rproc_mem_entry_init(&rproc->dev, NULL,
@@ -261,16 +265,19 @@ static int add_mem_regions_carveout(struct rproc *rproc)
 							 rmem->size, rmem->base,
 							 zynqmp_r5_mem_region_map,
 							 zynqmp_r5_mem_region_unmap,
-							 rmem->name);
+							 it.node->name);
 		}
 
-		if (!rproc_mem)
+		if (!rproc_mem) {
+			of_node_put(it.node);
 			return -ENOMEM;
+		}
 
 		rproc_add_carveout(rproc, rproc_mem);
 
 		dev_dbg(&rproc->dev, "reserved mem carveout %s addr=%llx, size=0x%llx",
-			rmem->name, rmem->base, rmem->size);
+			it.node->name, rmem->base, rmem->size);
+		i++;
 	}
 
 	return 0;
@@ -726,59 +733,6 @@ static int zynqmp_r5_get_tcm_node(struct zynqmp_r5_cluster *cluster)
 	return 0;
 }
 
-/**
- * zynqmp_r5_get_mem_region_node()
- * parse memory-region property and get reserved mem regions
- *
- * @r5_core: pointer to zynqmp_r5_core type object
- *
- * Return: 0 for success and error code for failure.
- */
-static int zynqmp_r5_get_mem_region_node(struct zynqmp_r5_core *r5_core)
-{
-	struct device_node *np, *rmem_np;
-	struct reserved_mem **rmem;
-	int res_mem_count, i;
-	struct device *dev;
-
-	dev = r5_core->dev;
-	np = r5_core->np;
-
-	res_mem_count = of_property_count_elems_of_size(np, "memory-region",
-							sizeof(phandle));
-	if (res_mem_count <= 0) {
-		dev_warn(dev, "failed to get memory-region property %d\n",
-			 res_mem_count);
-		return 0;
-	}
-
-	rmem = devm_kcalloc(dev, res_mem_count,
-			    sizeof(struct reserved_mem *), GFP_KERNEL);
-	if (!rmem)
-		return -ENOMEM;
-
-	for (i = 0; i < res_mem_count; i++) {
-		rmem_np = of_parse_phandle(np, "memory-region", i);
-		if (!rmem_np)
-			goto release_rmem;
-
-		rmem[i] = of_reserved_mem_lookup(rmem_np);
-		if (!rmem[i]) {
-			of_node_put(rmem_np);
-			goto release_rmem;
-		}
-
-		of_node_put(rmem_np);
-	}
-
-	r5_core->rmem_count = res_mem_count;
-	r5_core->rmem = rmem;
-	return 0;
-
-release_rmem:
-	return -EINVAL;
-}
-
 /*
  * zynqmp_r5_core_init()
  * Create and initialize zynqmp_r5_core type object
@@ -806,10 +760,6 @@ static int zynqmp_r5_core_init(struct zynqmp_r5_cluster *cluster,
 	for (i = 0; i < cluster->core_count; i++) {
 		r5_core = cluster->r5_cores[i];
 
-		ret = zynqmp_r5_get_mem_region_node(r5_core);
-		if (ret)
-			dev_warn(dev, "memory-region prop failed %d\n", ret);
-
 		/* Initialize r5 cores with power-domains parsed from dts */
 		ret = of_property_read_u32_index(r5_core->np, "power-domains",
 						 1, &r5_core->pm_domain_id);
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH v5 5/5] remoteproc: xilinx: add mailbox channels for rpmsg

[Tanmay Shah]

This patch makes each r5 core mailbox client and uses
tx and rx channels to send and receive data to/from
remote processor respectively. This is needed for rpmsg
communication to remote processor.

Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---

Changes in v5:
  - Remove spurious change
  - Include kick op regardless status of mailbox nodes in dts

Changes in v4:
  - move zynqmp_r5_setup_mbox from zynqmp_r5_add_rproc_core to
    zynqmp_r5_cluster_init

Changes in v3:
  - fix multi-line comment format
  - do not mixup mailbox information with memory-regions
  - fix redundant dev_warn for split mode
  - setting up mailboxes should return an error code
  - redesign driver to move mailbox setup during driver probe
  - add .kick function only if mailbox setup is success

v2: https://lore.kernel.org/all/20230126213154.1707300-1-tanmay.shah@amd.com/

 drivers/remoteproc/xlnx_r5_remoteproc.c | 218 +++++++++++++++++++++++-
 1 file changed, 217 insertions(+), 1 deletion(-)

diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
index 5dd007622603..53e3e4bd0f38 100644
--- a/drivers/remoteproc/xlnx_r5_remoteproc.c
+++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
@@ -8,16 +8,23 @@
 #include <linux/dma-mapping.h>
 #include <linux/firmware/xlnx-zynqmp.h>
 #include <linux/kernel.h>
+#include <linux/mailbox_client.h>
+#include <linux/mailbox/zynqmp-ipi-message.h>
 #include <linux/module.h>
 #include <linux/of_address.h>
 #include <linux/of_platform.h>
 #include <linux/of_reserved_mem.h>
 #include <linux/platform_device.h>
 #include <linux/remoteproc.h>
-#include <linux/slab.h>
 
 #include "remoteproc_internal.h"
 
+/* IPI buffer MAX length */
+#define IPI_BUF_LEN_MAX	32U
+
+/* RX mailbox client buffer max length */
+#define MBOX_CLIENT_BUF_MAX	(IPI_BUF_LEN_MAX + \
+				 sizeof(struct zynqmp_ipi_message))
 /*
  * settings for RPU cluster mode which
  * reflects possible values of xlnx,cluster-mode dt-property
@@ -43,6 +50,27 @@ struct mem_bank_data {
 	char *bank_name;
 };
 
+/**
+ * struct mbox_info
+ *
+ * @rx_mc_buf: to copy data from mailbox rx channel
+ * @tx_mc_buf: to copy data to mailbox tx channel
+ * @r5_core: this mailbox's corresponding r5_core pointer
+ * @mbox_work: schedule work after receiving data from mailbox
+ * @mbox_cl: mailbox client
+ * @tx_chan: mailbox tx channel
+ * @rx_chan: mailbox rx channel
+ */
+struct mbox_info {
+	unsigned char rx_mc_buf[MBOX_CLIENT_BUF_MAX];
+	unsigned char tx_mc_buf[MBOX_CLIENT_BUF_MAX];
+	struct zynqmp_r5_core *r5_core;
+	struct work_struct mbox_work;
+	struct mbox_client mbox_cl;
+	struct mbox_chan *tx_chan;
+	struct mbox_chan *rx_chan;
+};
+
 /*
  * Hardcoded TCM bank values. This will be removed once TCM bindings are
  * accepted for system-dt specifications and upstreamed in linux kernel
@@ -63,6 +91,7 @@ static const struct mem_bank_data zynqmp_tcm_banks[] = {
  * @tcm_banks: array of each TCM bank data
  * @rproc: rproc handle
  * @pm_domain_id: RPU CPU power domain id
+ * @ipi: pointer to mailbox information
  */
 struct zynqmp_r5_core {
 	struct device *dev;
@@ -71,6 +100,7 @@ struct zynqmp_r5_core {
 	struct mem_bank_data **tcm_banks;
 	struct rproc *rproc;
 	u32 pm_domain_id;
+	struct mbox_info *ipi;
 };
 
 /**
@@ -88,6 +118,178 @@ struct zynqmp_r5_cluster {
 	struct zynqmp_r5_core **r5_cores;
 };
 
+/**
+ * event_notified_idr_cb() - callback for vq_interrupt per notifyid
+ * @id: rproc->notify id
+ * @ptr: pointer to idr private data
+ * @data: data passed to idr_for_each callback
+ *
+ * Pass notification to remoteproc virtio
+ *
+ * Return: 0. having return is to satisfy the idr_for_each() function
+ *          pointer input argument requirement.
+ **/
+static int event_notified_idr_cb(int id, void *ptr, void *data)
+{
+	struct rproc *rproc = data;
+
+	if (rproc_vq_interrupt(rproc, id) == IRQ_NONE)
+		dev_dbg(&rproc->dev, "data not found for vqid=%d\n", id);
+
+	return 0;
+}
+
+/**
+ * handle_event_notified() - remoteproc notification work function
+ * @work: pointer to the work structure
+ *
+ * It checks each registered remoteproc notify IDs.
+ */
+static void handle_event_notified(struct work_struct *work)
+{
+	struct mbox_info *ipi;
+	struct rproc *rproc;
+
+	ipi = container_of(work, struct mbox_info, mbox_work);
+	rproc = ipi->r5_core->rproc;
+
+	/*
+	 * We only use IPI for interrupt. The RPU firmware side may or may
+	 * not write the notifyid when it trigger IPI.
+	 * And thus, we scan through all the registered notifyids and
+	 * find which one is valid to get the message.
+	 * Even if message from firmware is NULL, we attempt to get vqid
+	 */
+	idr_for_each(&rproc->notifyids, event_notified_idr_cb, rproc);
+}
+
+/**
+ * zynqmp_r5_mb_rx_cb() - receive channel mailbox callback
+ * @cl: mailbox client
+ * @msg: message pointer
+ *
+ * Receive data from ipi buffer, ack interrupt and then
+ * it will schedule the R5 notification work.
+ */
+static void zynqmp_r5_mb_rx_cb(struct mbox_client *cl, void *msg)
+{
+	struct zynqmp_ipi_message *ipi_msg, *buf_msg;
+	struct mbox_info *ipi;
+	size_t len;
+
+	ipi = container_of(cl, struct mbox_info, mbox_cl);
+
+	/* copy data from ipi buffer to r5_core */
+	ipi_msg = (struct zynqmp_ipi_message *)msg;
+	buf_msg = (struct zynqmp_ipi_message *)ipi->rx_mc_buf;
+	len = ipi_msg->len;
+	if (len > IPI_BUF_LEN_MAX) {
+		dev_warn(cl->dev, "msg size exceeded than %d\n",
+			 IPI_BUF_LEN_MAX);
+		len = IPI_BUF_LEN_MAX;
+	}
+	buf_msg->len = len;
+	memcpy(buf_msg->data, ipi_msg->data, len);
+
+	/* received and processed interrupt ack */
+	if (mbox_send_message(ipi->rx_chan, NULL) < 0)
+		dev_err(cl->dev, "ack failed to mbox rx_chan\n");
+
+	schedule_work(&ipi->mbox_work);
+}
+
+/**
+ * zynqmp_r5_setup_mbox() - Setup mailboxes related properties
+ *			    this is used for each individual R5 core
+ *
+ * @cdev: child node device
+ *
+ * Function to setup mailboxes related properties
+ * return : NULL if failed else pointer to mbox_info
+ */
+static struct mbox_info *zynqmp_r5_setup_mbox(struct device *cdev)
+{
+	struct mbox_client *mbox_cl;
+	struct mbox_info *ipi;
+
+	ipi = kzalloc(sizeof(*ipi), GFP_KERNEL);
+	if (!ipi)
+		return NULL;
+
+	mbox_cl = &ipi->mbox_cl;
+	mbox_cl->rx_callback = zynqmp_r5_mb_rx_cb;
+	mbox_cl->tx_block = false;
+	mbox_cl->knows_txdone = false;
+	mbox_cl->tx_done = NULL;
+	mbox_cl->dev = cdev;
+
+	/* Request TX and RX channels */
+	ipi->tx_chan = mbox_request_channel_byname(mbox_cl, "tx");
+	if (IS_ERR(ipi->tx_chan)) {
+		ipi->tx_chan = NULL;
+		kfree(ipi);
+		dev_warn(cdev, "mbox tx channel request failed\n");
+		return NULL;
+	}
+
+	ipi->rx_chan = mbox_request_channel_byname(mbox_cl, "rx");
+	if (IS_ERR(ipi->rx_chan)) {
+		mbox_free_channel(ipi->tx_chan);
+		ipi->rx_chan = NULL;
+		ipi->tx_chan = NULL;
+		kfree(ipi);
+		dev_warn(cdev, "mbox rx channel request failed\n");
+		return NULL;
+	}
+
+	INIT_WORK(&ipi->mbox_work, handle_event_notified);
+
+	return ipi;
+}
+
+static void zynqmp_r5_free_mbox(struct mbox_info *ipi)
+{
+	if (!ipi)
+		return;
+
+	if (ipi->tx_chan) {
+		mbox_free_channel(ipi->tx_chan);
+		ipi->tx_chan = NULL;
+	}
+
+	if (ipi->rx_chan) {
+		mbox_free_channel(ipi->rx_chan);
+		ipi->rx_chan = NULL;
+	}
+
+	kfree(ipi);
+}
+
+/*
+ * zynqmp_r5_core_kick() - kick a firmware if mbox is provided
+ * @rproc: r5 core's corresponding rproc structure
+ * @vqid: virtqueue ID
+ */
+static void zynqmp_r5_rproc_kick(struct rproc *rproc, int vqid)
+{
+	struct zynqmp_r5_core *r5_core = rproc->priv;
+	struct device *dev = r5_core->dev;
+	struct zynqmp_ipi_message *mb_msg;
+	struct mbox_info *ipi;
+	int ret;
+
+	ipi = r5_core->ipi;
+	if (!ipi)
+		return;
+
+	mb_msg = (struct zynqmp_ipi_message *)ipi->tx_mc_buf;
+	memcpy(mb_msg->data, &vqid, sizeof(vqid));
+	mb_msg->len = sizeof(vqid);
+	ret = mbox_send_message(ipi->tx_chan, mb_msg);
+	if (ret < 0)
+		dev_warn(dev, "failed to send message\n");
+}
+
 /*
  * zynqmp_r5_set_mode()
  *
@@ -624,6 +826,7 @@ static const struct rproc_ops zynqmp_r5_rproc_ops = {
 	.find_loaded_rsc_table = rproc_elf_find_loaded_rsc_table,
 	.sanity_check	= rproc_elf_sanity_check,
 	.get_boot_addr	= rproc_elf_get_boot_addr,
+	.kick		= zynqmp_r5_rproc_kick,
 };
 
 /**
@@ -799,6 +1002,7 @@ static int zynqmp_r5_cluster_init(struct zynqmp_r5_cluster *cluster)
 	struct device_node *child;
 	enum rpu_tcm_comb tcm_mode;
 	int core_count, ret, i;
+	struct mbox_info *ipi;
 
 	ret = of_property_read_u32(dev_node, "xlnx,cluster-mode", &cluster_mode);
 
@@ -878,6 +1082,16 @@ static int zynqmp_r5_cluster_init(struct zynqmp_r5_cluster *cluster)
 			goto release_r5_cores;
 		}
 
+		/*
+		 * If mailbox nodes are disabled using "status" property then
+		 * setting up mailbox channels will fail.
+		 */
+		ipi = zynqmp_r5_setup_mbox(&child_pdev->dev);
+		if (ipi) {
+			r5_cores[i]->ipi = ipi;
+			ipi->r5_core = r5_cores[i];
+		}
+
 		/*
 		 * If two child nodes are available in dts in lockstep mode,
 		 * then ignore second child node.
@@ -915,6 +1129,7 @@ static int zynqmp_r5_cluster_init(struct zynqmp_r5_cluster *cluster)
 	while (i >= 0) {
 		put_device(child_devs[i]);
 		if (r5_cores[i]) {
+			zynqmp_r5_free_mbox(r5_cores[i]->ipi);
 			of_reserved_mem_device_release(r5_cores[i]->dev);
 			rproc_del(r5_cores[i]->rproc);
 			rproc_free(r5_cores[i]->rproc);
@@ -939,6 +1154,7 @@ static void zynqmp_r5_cluster_exit(void *data)
 
 	for (i = 0; i < cluster->core_count; i++) {
 		r5_core = cluster->r5_cores[i];
+		zynqmp_r5_free_mbox(r5_core->ipi);
 		of_reserved_mem_device_release(r5_core->dev);
 		put_device(r5_core->dev);
 		rproc_del(r5_core->rproc);
-- 
2.25.1


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH v5 0/5] drivers: remoteproc: xilinx: add mailbox support

[Mathieu Poirier]

On Fri, Mar 10, 2023 at 05:24:03PM -0800, Tanmay Shah wrote:
> This series adds mailbox support required for rpmsg communication
> in Xilinx remoteproc driver. It also fixes few bugs in Xilinx
> zynqmp-ipi driver that handles corner cases.
> 
> ---
> 
> Changes in v5:
>   - mailbox node count should not be 0 for driver to probe
>   - Remove spurious change
>   - Include kick op regardless status of mailbox nodes in dts
> 
> Changes in v4:
>   - split mailbox driver patch into 3 different patches
>   - use of_phandle_iterator_next to get memory-region nodes
>   - move zynqmp_r5_setup_mbox from zynqmp_r5_add_rproc_core to
>     zynqmp_r5_cluster_init
> 
> Changes in v3:
>   - split single patch into multiple patches
>   - add appropriate fixes tag for patches fixing previous bug in the driver
>   - make enhancements in zynqmp-ipi driver that handles corener cases
>   - fix memory-region carveout names
>   - fix multi-line comment format
>   - do not mixup mailbox information with memory-regions
>   - fix redundant dev_warn for split mode
>   - Setting up mailboxes should return an error code
>   - Move mailbox setup during driver probe
>   - Add .kick function only if mailbox setup is success
> 
> v2: https://lore.kernel.org/all/20230126213154.1707300-1-tanmay.shah@amd.com/
> 
> 
> Tanmay Shah (5):
>   mailbox: zynqmp: fix counts of child nodes
>   mailbox: zynqmp: fix IPI isr handling
>   mailbox: zynqmp: fix typo in IPI documentation
>   drivers: remoteproc: xilinx: fix carveout names
>   remoteproc: xilinx: add mailbox channels for rpmsg
> 
>  drivers/mailbox/zynqmp-ipi-mailbox.c       |  13 +-
>  drivers/remoteproc/xlnx_r5_remoteproc.c    | 308 ++++++++++++++++-----
>  include/linux/mailbox/zynqmp-ipi-message.h |   2 +-
>  3 files changed, 247 insertions(+), 76 deletions(-)
>

I have applied this set.

Thanks,
Mathieu

> 
> base-commit: fe15c26ee26efa11741a7b632e9f23b01aca4cc6
> -- 
> 2.25.1
> 

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel