[PATCH 0/3] remoteproc: xlnx: enhancements and new features

[PATCH 0/3] remoteproc: xlnx: enhancements and new features

[Tanmay Shah]

Introduce new enhancements in the remoteproc xlnx driver. Handle mailbox
optional properties gracefully. Handle buffer-less IPI without crash i.e.
no message is expected from the mailbox callback. Also gracefully free
mailbox channels during shutdown callback.
These small enhancements make driver more robust.

Ben Levinsky (1):
  remoteproc: xlnx: Only access buffer information if IPI is buffered

Tanmay Shah (2):
  remoteproc: xlnx: avoid mailbox setup
  remoteproc: zynqmp: release mailbox channels on shutdown

 drivers/remoteproc/xlnx_r5_remoteproc.c | 26 ++++++++++++++++---------
 1 file changed, 17 insertions(+), 9 deletions(-)


base-commit: 098493c6dced7b02545e8bd0053ef4099a2b769e
-- 
2.34.1

[PATCH 1/3] remoteproc: xlnx: avoid mailbox setup

[Tanmay Shah]

Mailbox properties are optional in the remoteproc xlnx bindings.
If mailbox properties are not found in device-tree it's not fatal error
in the driver. Driver will print warning messages and continue with
normal operation. However, these warning messages can be interpreted as
error. Check "mboxes" and "mbox-names" properties in the driver and setup
mailbox only if they are available in the device-tree.
This will avoid any unwanted warning/error messages.

Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---
 drivers/remoteproc/xlnx_r5_remoteproc.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
index 5a468d959f1e..148d8c622566 100644
--- a/drivers/remoteproc/xlnx_r5_remoteproc.c
+++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
@@ -265,6 +265,10 @@ static struct mbox_info *zynqmp_r5_setup_mbox(struct device *cdev)
 	struct mbox_client *mbox_cl;
 	struct mbox_info *ipi;
 
+	if (!of_property_present(dev_of_node(cdev), "mboxes") ||
+	    !of_property_present(dev_of_node(cdev), "mbox-names"))
+		return NULL;
+
 	ipi = kzalloc_obj(*ipi);
 	if (!ipi)
 		return NULL;
-- 
2.34.1

[PATCH 2/3] remoteproc: xlnx: Only access buffer information if IPI is buffered

[Tanmay Shah]

From: Ben Levinsky <ben.levinsky@amd.com>

In the receive callback check if message is NULL to prevent
possibility of crash by NULL pointer dereferencing.

Signed-off-by: Ben Levinsky <ben.levinsky@amd.com>
Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---
 drivers/remoteproc/xlnx_r5_remoteproc.c | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
index 148d8c622566..5e92dc51f1c0 100644
--- a/drivers/remoteproc/xlnx_r5_remoteproc.c
+++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
@@ -232,17 +232,19 @@ static void zynqmp_r5_mb_rx_cb(struct mbox_client *cl, void *msg)
 
 	ipi = container_of(cl, struct mbox_info, mbox_cl);
 
-	/* copy data from ipi buffer to r5_core */
+	/* copy data from ipi buffer to r5_core if IPI is buffered. */
 	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;
+	if (ipi_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);
 	}
-	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)
-- 
2.34.1

[PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Tanmay Shah]

mailbox driver can't introduce shutdown callback, as it might endup
closing mbox channels prematurely. By allowing the client driver to
manage the shutdown process, it's ensured that mailbox channels are
closed only when they are no longer needed.

Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
---
 drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
index 5e92dc51f1c0..50a9974f3202 100644
--- a/drivers/remoteproc/xlnx_r5_remoteproc.c
+++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
@@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
 			dev_err(cluster->dev, "failed to %s rproc %d\n",
 				rproc_state_str, rproc->index);
 		}
+
+		zynqmp_r5_free_mbox(r5_core->ipi);
 	}
 }
 
-- 
2.34.1

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Mathieu Poirier]

On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
> mailbox driver can't introduce shutdown callback, as it might endup
> closing mbox channels prematurely. By allowing the client driver to
> manage the shutdown process, it's ensured that mailbox channels are
> closed only when they are no longer needed.
> 
> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> ---
>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
> index 5e92dc51f1c0..50a9974f3202 100644
> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
>  				rproc_state_str, rproc->index);
>  		}
> +
> +		zynqmp_r5_free_mbox(r5_core->ipi);

This is already called in zynqmp_r5_cluster_exit(), why doing it here again?

I have applied the other two patches in this series.

Thanks,
Mathieu

>  	}
>  }
>  
> -- 
> 2.34.1
> 

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Shah, Tanmay]

On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
> On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
>> mailbox driver can't introduce shutdown callback, as it might endup
>> closing mbox channels prematurely. By allowing the client driver to
>> manage the shutdown process, it's ensured that mailbox channels are
>> closed only when they are no longer needed.
>>
>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
>> ---
>>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
>> index 5e92dc51f1c0..50a9974f3202 100644
>> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
>> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
>> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
>>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
>>  				rproc_state_str, rproc->index);
>>  		}
>> +
>> +		zynqmp_r5_free_mbox(r5_core->ipi);
> 
> This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
> 

Hi,

Thanks for reviews.
I think cluster_exit() call is called only during driver unload. Where
as shutdown callback is called during power-off commands of linux like
reboot or shutdown.

That is why I am calling it separately during shutdown() callback.

Thanks,
Tanmay

> I have applied the other two patches in this series.
> 
> Thanks,
> Mathieu
> 
>>  	}
>>  }
>>  
>> -- 
>> 2.34.1
>>

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Mathieu Poirier]

On Tue, Mar 10, 2026 at 11:04:54AM -0500, Shah, Tanmay wrote:
> 
> 
> On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
> > On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
> >> mailbox driver can't introduce shutdown callback, as it might endup
> >> closing mbox channels prematurely. By allowing the client driver to
> >> manage the shutdown process, it's ensured that mailbox channels are
> >> closed only when they are no longer needed.
> >>
> >> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> >> ---
> >>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
> >>  1 file changed, 2 insertions(+)
> >>
> >> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> index 5e92dc51f1c0..50a9974f3202 100644
> >> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
> >>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
> >>  				rproc_state_str, rproc->index);
> >>  		}
> >> +
> >> +		zynqmp_r5_free_mbox(r5_core->ipi);
> > 
> > This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
> > 
> 
> Hi,
> 
> Thanks for reviews.
> I think cluster_exit() call is called only during driver unload. Where
> as shutdown callback is called during power-off commands of linux like
> reboot or shutdown.
> 
> That is why I am calling it separately during shutdown() callback.

The problem here is that zynqmp_r5_free_mbox() is called twice.  At shutdown
time all drivers go throuth the normal driver shutdown process where
platform_driver.shutdown() is called and subsequently device management
callbacks such as zynqmp_r5_cluster_exit().  The same applies to loading and
unloading of drivers.  Unless there is a corner case I can't see, calling
zynqmp_r5_free_mbox() twice is not needed.

> 
> Thanks,
> Tanmay
> 
> > I have applied the other two patches in this series.
> > 
> > Thanks,
> > Mathieu
> > 
> >>  	}
> >>  }
> >>  
> >> -- 
> >> 2.34.1
> >>
> 

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Shah, Tanmay]

On 3/11/2026 12:12 PM, Mathieu Poirier wrote:
> On Tue, Mar 10, 2026 at 11:04:54AM -0500, Shah, Tanmay wrote:
>>
>>
>> On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
>>> On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
>>>> mailbox driver can't introduce shutdown callback, as it might endup
>>>> closing mbox channels prematurely. By allowing the client driver to
>>>> manage the shutdown process, it's ensured that mailbox channels are
>>>> closed only when they are no longer needed.
>>>>
>>>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
>>>> ---
>>>>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
>>>>  1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> index 5e92dc51f1c0..50a9974f3202 100644
>>>> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
>>>>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
>>>>  				rproc_state_str, rproc->index);
>>>>  		}
>>>> +
>>>> +		zynqmp_r5_free_mbox(r5_core->ipi);
>>>
>>> This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
>>>
>>
>> Hi,
>>
>> Thanks for reviews.
>> I think cluster_exit() call is called only during driver unload. Where
>> as shutdown callback is called during power-off commands of linux like
>> reboot or shutdown.
>>
>> That is why I am calling it separately during shutdown() callback.
> 
> The problem here is that zynqmp_r5_free_mbox() is called twice.  At shutdown
> time all drivers go throuth the normal driver shutdown process where
> platform_driver.shutdown() is called and subsequently device management
> callbacks such as zynqmp_r5_cluster_exit().  The same applies to loading and
> unloading of drivers.  Unless there is a corner case I can't see, calling
> zynqmp_r5_free_mbox() twice is not needed.
> 

Hi Mathieu,

I tested this patch again. On my platform I don't see
zynqmp_r5_cluster_exit() being called when issuing `reboot` command.

It only calls platform_driver.shutdown() callback.

I think device management APIs are called only during driver remove
callback, and not shutdown sequence.

Thanks,
Tanmay

>>
>> Thanks,
>> Tanmay
>>
>>> I have applied the other two patches in this series.
>>>
>>> Thanks,
>>> Mathieu
>>>
>>>>  	}
>>>>  }
>>>>  
>>>> -- 
>>>> 2.34.1
>>>>
>>

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Mathieu Poirier]

On Tue, Mar 10, 2026 at 11:04:54AM -0500, Shah, Tanmay wrote:
> 
> 
> On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
> > On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
> >> mailbox driver can't introduce shutdown callback, as it might endup
> >> closing mbox channels prematurely. By allowing the client driver to
> >> manage the shutdown process, it's ensured that mailbox channels are
> >> closed only when they are no longer needed.
> >>
> >> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> >> ---
> >>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
> >>  1 file changed, 2 insertions(+)
> >>
> >> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> index 5e92dc51f1c0..50a9974f3202 100644
> >> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
> >>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
> >>  				rproc_state_str, rproc->index);
> >>  		}
> >> +
> >> +		zynqmp_r5_free_mbox(r5_core->ipi);
> > 
> > This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
> > 
> 
> Hi,
> 
> Thanks for reviews.
> I think cluster_exit() call is called only during driver unload. Where
> as shutdown callback is called during power-off commands of linux like
> reboot or shutdown.
> 
> That is why I am calling it separately during shutdown() callback.

Then call zynqmp_r5_free_mbox() from zynqmp_r5_remoteproc_shutdown() rather than
zynqmp_r5_cluster_exit().

> 
> Thanks,
> Tanmay
> 
> > I have applied the other two patches in this series.
> > 
> > Thanks,
> > Mathieu
> > 
> >>  	}
> >>  }
> >>  
> >> -- 
> >> 2.34.1
> >>
> 

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Shah, Tanmay]

On 3/16/2026 10:38 AM, Mathieu Poirier wrote:
> On Tue, Mar 10, 2026 at 11:04:54AM -0500, Shah, Tanmay wrote:
>>
>>
>> On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
>>> On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
>>>> mailbox driver can't introduce shutdown callback, as it might endup
>>>> closing mbox channels prematurely. By allowing the client driver to
>>>> manage the shutdown process, it's ensured that mailbox channels are
>>>> closed only when they are no longer needed.
>>>>
>>>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
>>>> ---
>>>>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
>>>>  1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> index 5e92dc51f1c0..50a9974f3202 100644
>>>> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
>>>> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
>>>>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
>>>>  				rproc_state_str, rproc->index);
>>>>  		}
>>>> +
>>>> +		zynqmp_r5_free_mbox(r5_core->ipi);
>>>
>>> This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
>>>
>>
>> Hi,
>>
>> Thanks for reviews.
>> I think cluster_exit() call is called only during driver unload. Where
>> as shutdown callback is called during power-off commands of linux like
>> reboot or shutdown.
>>
>> That is why I am calling it separately during shutdown() callback.
> 
> Then call zynqmp_r5_free_mbox() from zynqmp_r5_remoteproc_shutdown() rather than
> zynqmp_r5_cluster_exit().
> 

I have to call zynqmp_r5_free_mbox() from both handlers i.e.
platform_driver.remove() and platform_driver.shutdown(). I can't skip
either of them.

Because during driver unload (rmmod), shutdown() handler won't be
called. And during power-off commands (reboot, shutdown) remove()
handler won't be called.

Let me know if I am still missing something.

Thanks,
Tanmay

>>
>> Thanks,
>> Tanmay
>>
>>> I have applied the other two patches in this series.
>>>
>>> Thanks,
>>> Mathieu
>>>
>>>>  	}
>>>>  }
>>>>  
>>>> -- 
>>>> 2.34.1
>>>>
>>

Re: [PATCH 3/3] remoteproc: zynqmp: release mailbox channels on shutdown

[Mathieu Poirier]

On Mon, Mar 16, 2026 at 11:37:05AM -0500, Shah, Tanmay wrote:
> 
> 
> On 3/16/2026 10:38 AM, Mathieu Poirier wrote:
> > On Tue, Mar 10, 2026 at 11:04:54AM -0500, Shah, Tanmay wrote:
> >>
> >>
> >> On 3/10/2026 10:29 AM, Mathieu Poirier wrote:
> >>> On Tue, Mar 03, 2026 at 03:51:28PM -0800, Tanmay Shah wrote:
> >>>> mailbox driver can't introduce shutdown callback, as it might endup
> >>>> closing mbox channels prematurely. By allowing the client driver to
> >>>> manage the shutdown process, it's ensured that mailbox channels are
> >>>> closed only when they are no longer needed.
> >>>>
> >>>> Signed-off-by: Tanmay Shah <tanmay.shah@amd.com>
> >>>> ---
> >>>>  drivers/remoteproc/xlnx_r5_remoteproc.c | 2 ++
> >>>>  1 file changed, 2 insertions(+)
> >>>>
> >>>> diff --git a/drivers/remoteproc/xlnx_r5_remoteproc.c b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >>>> index 5e92dc51f1c0..50a9974f3202 100644
> >>>> --- a/drivers/remoteproc/xlnx_r5_remoteproc.c
> >>>> +++ b/drivers/remoteproc/xlnx_r5_remoteproc.c
> >>>> @@ -1490,6 +1490,8 @@ static void zynqmp_r5_remoteproc_shutdown(struct platform_device *pdev)
> >>>>  			dev_err(cluster->dev, "failed to %s rproc %d\n",
> >>>>  				rproc_state_str, rproc->index);
> >>>>  		}
> >>>> +
> >>>> +		zynqmp_r5_free_mbox(r5_core->ipi);
> >>>
> >>> This is already called in zynqmp_r5_cluster_exit(), why doing it here again?
> >>>
> >>
> >> Hi,
> >>
> >> Thanks for reviews.
> >> I think cluster_exit() call is called only during driver unload. Where
> >> as shutdown callback is called during power-off commands of linux like
> >> reboot or shutdown.
> >>
> >> That is why I am calling it separately during shutdown() callback.
> > 
> > Then call zynqmp_r5_free_mbox() from zynqmp_r5_remoteproc_shutdown() rather than
> > zynqmp_r5_cluster_exit().
> > 
> 
> I have to call zynqmp_r5_free_mbox() from both handlers i.e.
> platform_driver.remove() and platform_driver.shutdown(). I can't skip
> either of them.
> 
> Because during driver unload (rmmod), shutdown() handler won't be
> called. And during power-off commands (reboot, shutdown) remove()
> handler won't be called.

I was under the impression that during system shutdown, ->remove() and
->shutdown() were called sequentially.  I looked into the platform_driver
structure documentation and nothing of that nature is described, leading me to
beleive your assessment is correct.

I'll apply your patch.

> 
> Let me know if I am still missing something.
> 
> Thanks,
> Tanmay
> 
> >>
> >> Thanks,
> >> Tanmay
> >>
> >>> I have applied the other two patches in this series.
> >>>
> >>> Thanks,
> >>> Mathieu
> >>>
> >>>>  	}
> >>>>  }
> >>>>  
> >>>> -- 
> >>>> 2.34.1
> >>>>
> >>
>