[PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode

Nandor Han posted 2 patches 1 month, 1 week ago
There is a newer version of this series
[PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Nandor Han 1 month, 1 week ago
Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
running in low power modes (STOP and DOZE). By default, the watchdog is
suspended in these modes. This property provides the option to keep the
watchdog active during low power states when needed.

Signed-off-by: Nandor Han <nandor.han@gehealthcare.com>
---
 Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml b/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
index 34951783a633..4dd9cc81c926 100644
--- a/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
@@ -64,6 +64,12 @@ properties:
       If present, the watchdog device is suspended in WAIT mode
       (Suspend-to-Idle). Only supported on certain devices.
 
+  fsl,wdt-continue-in-low-power:
+    $ref: /schemas/types.yaml#/definitions/flag
+    description: |
+      If present, the watchdog device continues to run in low power modes
+      (STOP and DOZE). By default, the watchdog is suspended in these modes.
+
 required:
   - compatible
   - interrupts
-- 
2.39.5
Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Peng Fan 1 month ago
Hi Nandor,

On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>running in low power modes (STOP and DOZE). By default, the watchdog is
>suspended in these modes. This property provides the option to keep the
>watchdog active during low power states when needed.

From the cover letter, I see that system should reset if runs into
low power mode. So your system are not allowed to run into low power mode.

If my understanding is correct, DT property should not be used.
You may need to use sysfs or module_parameter for your case.

Regards,
Peng

>
>Signed-off-by: Nandor Han <nandor.han@gehealthcare.com>
>---
> Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml | 6 ++++++
> 1 file changed, 6 insertions(+)
>
>diff --git a/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml b/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
>index 34951783a633..4dd9cc81c926 100644
>--- a/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
>+++ b/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
>@@ -64,6 +64,12 @@ properties:
>       If present, the watchdog device is suspended in WAIT mode
>       (Suspend-to-Idle). Only supported on certain devices.
> 
>+  fsl,wdt-continue-in-low-power:
>+    $ref: /schemas/types.yaml#/definitions/flag
>+    description: |
>+      If present, the watchdog device continues to run in low power modes
>+      (STOP and DOZE). By default, the watchdog is suspended in these modes.
>+
> required:
>   - compatible
>   - interrupts
>-- 
>2.39.5
>
Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Nandor Han 1 month ago
On 1/6/26 13:51, Peng Fan wrote:
> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>
> Hi Nandor,
>
> On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>> running in low power modes (STOP and DOZE). By default, the watchdog is
>> suspended in these modes. This property provides the option to keep the
>> watchdog active during low power states when needed.
> >From the cover letter, I see that system should reset if runs into
> low power mode. So your system are not allowed to run into low power mode.
>
> If my understanding is correct, DT property should not be used.
> You may need to use sysfs or module_parameter for your case.

Thanks Peng.

Yes, you're correct, our system should not run in low power mode.

My implementation is based on the fact that the driver provides already a Device Tree configuration for the WAIT low power mode.
To maintainconsistency, the patch simply extends support to configure the watchdog during STOP and DOZE power modes.

One important consideration is that the watchdog must be fully configured during boot to catch any potential issues early.
Since this is a hardware property of the watchdog, we considered that device tree is a good place for this configuration.

Regards,
Nandor



Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Krzysztof Kozlowski 1 month, 1 week ago
On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
> running in low power modes (STOP and DOZE). By default, the watchdog is
> suspended in these modes. This property provides the option to keep the
> watchdog active during low power states when needed.

And why exactly would that be a DT property? If system is sleeping
(assuming this is what you meant by low power), no one will pet the dog,
thus watchdog makes no sense.

Otherwise I fail to see how this is a hardware property and we do not
accept SW properties (see writing bindings, numerous presentations).

Best regards,
Krzysztof
Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Nandor Han 1 month ago
On 12/30/25 14:34, Krzysztof Kozlowski wrote:
> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>
> On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>> running in low power modes (STOP and DOZE). By default, the watchdog is
>> suspended in these modes. This property provides the option to keep the
>> watchdog active during low power states when needed.
> And why exactly would that be a DT property? If system is sleeping
> (assuming this is what you meant by low power), no one will pet the dog,
> thus watchdog makes no sense.

Thanks for the feedback Krzysztof and Guenter.

In our case, low-power mode is disabled. However, we have identified that under certain conditions,
specifically during simulated high-load scenarios, the device becomes unresponsive because it enters
one of these power states.

> Otherwise I fail to see how this is a hardware property and we do not
> accept SW properties (see writing bindings, numerous presentations).

Our system is based on the i.MX7D CPU and the watchdog peripheral supports the configuration:

(From i.MX 7Dual Applications Processor Reference Manual, Rev. 1, 01/2018, page: 1174)
---
WDZST
Watchdog Low Power. Determines the operation of the WDOG during low-power modes. This bit is write
once-only.
---
Given that our system does not support low-power modes, we intend to enable the watchdog across all power
states to ensure the device can recover properly under these conditions.

Regards,
Nandor

Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Krzysztof Kozlowski 1 month ago
On 07/01/2026 10:12, Nandor Han wrote:
> 
> On 12/30/25 14:34, Krzysztof Kozlowski wrote:
>> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>>
>> On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>>> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>>> running in low power modes (STOP and DOZE). By default, the watchdog is
>>> suspended in these modes. This property provides the option to keep the
>>> watchdog active during low power states when needed.
>> And why exactly would that be a DT property? If system is sleeping
>> (assuming this is what you meant by low power), no one will pet the dog,
>> thus watchdog makes no sense.
> 
> Thanks for the feedback Krzysztof and Guenter.
> 
> In our case, low-power mode is disabled. However, we have identified that under certain conditions,

If your system has low power mode disabled, then you do not need this
property - you already know that watchdog must continue (or whatever you
want to achieve here).

> specifically during simulated high-load scenarios, the device becomes unresponsive because it enters
> one of these power states.

Device as watchdog? I really do not understand your explanations, but
for sure system load is not relevant to DT property.

> 
>> Otherwise I fail to see how this is a hardware property and we do not
>> accept SW properties (see writing bindings, numerous presentations).
> 
> Our system is based on the i.MX7D CPU and the watchdog peripheral supports the configuration:
> 
> (From i.MX 7Dual Applications Processor Reference Manual, Rev. 1, 01/2018, page: 1174)
> ---
> WDZST
> Watchdog Low Power. Determines the operation of the WDOG during low-power modes. This bit is write
> once-only.
> ---
> Given that our system does not support low-power modes, we intend to enable the watchdog across all power
> states to ensure the device can recover properly under these conditions.

That's not what your property said. Your property said watchdog should
continue in low power modes. So when system enters low power mode, how
the watchdog petting would work?

Now you claim you want to enable it in low power mode but you do not
have low power mode? Does not make sense to me at all.

Best regards,
Krzysztof
Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Nandor Han 1 month ago
On 1/7/26 12:48, Krzysztof Kozlowski wrote:
> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>
> On 07/01/2026 10:12, Nandor Han wrote:
>> On 12/30/25 14:34, Krzysztof Kozlowski wrote:
>>> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>>>
>>> On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>>>> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>>>> running in low power modes (STOP and DOZE). By default, the watchdog is
>>>> suspended in these modes. This property provides the option to keep the
>>>> watchdog active during low power states when needed.
>>> And why exactly would that be a DT property? If system is sleeping
>>> (assuming this is what you meant by low power), no one will pet the dog,
>>> thus watchdog makes no sense.
>> Thanks for the feedback Krzysztof and Guenter.
>>
>> In our case, low-power mode is disabled. However, we have identified that under certain conditions,
> If your system has low power mode disabled, then you do not need this
> property - you already know that watchdog must continue (or whatever you
> want to achieve here).
>
>> specifically during simulated high-load scenarios, the device becomes unresponsive because it enters
>> one of these power states.
> Device as watchdog? I really do not understand your explanations, but
> for sure system load is not relevant to DT property.
>
>>> Otherwise I fail to see how this is a hardware property and we do not
>>> accept SW properties (see writing bindings, numerous presentations).
>> Our system is based on the i.MX7D CPU and the watchdog peripheral supports the configuration:
>>
>> (From i.MX 7Dual Applications Processor Reference Manual, Rev. 1, 01/2018, page: 1174)
>> ---
>> WDZST
>> Watchdog Low Power. Determines the operation of the WDOG during low-power modes. This bit is write
>> once-only.
>> ---
>> Given that our system does not support low-power modes, we intend to enable the watchdog across all power
>> states to ensure the device can recover properly under these conditions.
> That's not what your property said. Your property said watchdog should
> continue in low power modes. So when system enters low power mode, how
> the watchdog petting would work?
>
> Now you claim you want to enable it in low power mode but you do not
> have low power mode? Does not make sense to me at all.
>
> Best regards,
> Krzysztof
>
Sorry if anything is unclear. I would try to explain the change from the driver's point of view.

According to i.MX7D Reference Manual, the watchdog controller allows, via the WDOGx_WCR register, control over whether the watchdog continues or suspends in:

a)WAIT power state (bit 7: WDW)
b)STOP and DOSE power state(bit 0: WDZST).

The current driver implementation provides a Device Tree binding `fsl,suspend-in-wait` for configuring case (a) and forces the watchdog to be suspended in case (b).

My patch adds the ability to configure case (b) as well.

Regards,
Nandor



Re: [PATCH 2/2] dt-bindings: watchdog: fsl-imx: document continue in low power mode
Posted by Krzysztof Kozlowski 4 weeks ago
On 08/01/2026 08:58, Nandor Han wrote:
> 
> On 1/7/26 12:48, Krzysztof Kozlowski wrote:
>> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>>
>> On 07/01/2026 10:12, Nandor Han wrote:
>>> On 12/30/25 14:34, Krzysztof Kozlowski wrote:
>>>> CAUTION: This email originated from outside of GE HealthCare. Only open links or attachments if you trust the sender. Report suspicious emails using Outlook’s “Report” button.
>>>>
>>>> On Mon, Dec 29, 2025 at 04:50:00PM +0200, Nandor Han wrote:
>>>>> Property "fsl,wdt-continue-in-low-power" allows the watchdog to continue
>>>>> running in low power modes (STOP and DOZE). By default, the watchdog is
>>>>> suspended in these modes. This property provides the option to keep the
>>>>> watchdog active during low power states when needed.
>>>> And why exactly would that be a DT property? If system is sleeping
>>>> (assuming this is what you meant by low power), no one will pet the dog,
>>>> thus watchdog makes no sense.
>>> Thanks for the feedback Krzysztof and Guenter.
>>>
>>> In our case, low-power mode is disabled. However, we have identified that under certain conditions,
>> If your system has low power mode disabled, then you do not need this
>> property - you already know that watchdog must continue (or whatever you
>> want to achieve here).
>>
>>> specifically during simulated high-load scenarios, the device becomes unresponsive because it enters
>>> one of these power states.
>> Device as watchdog? I really do not understand your explanations, but
>> for sure system load is not relevant to DT property.
>>
>>>> Otherwise I fail to see how this is a hardware property and we do not
>>>> accept SW properties (see writing bindings, numerous presentations).
>>> Our system is based on the i.MX7D CPU and the watchdog peripheral supports the configuration:
>>>
>>> (From i.MX 7Dual Applications Processor Reference Manual, Rev. 1, 01/2018, page: 1174)
>>> ---
>>> WDZST
>>> Watchdog Low Power. Determines the operation of the WDOG during low-power modes. This bit is write
>>> once-only.
>>> ---
>>> Given that our system does not support low-power modes, we intend to enable the watchdog across all power
>>> states to ensure the device can recover properly under these conditions.
>> That's not what your property said. Your property said watchdog should
>> continue in low power modes. So when system enters low power mode, how
>> the watchdog petting would work?
>>
>> Now you claim you want to enable it in low power mode but you do not
>> have low power mode? Does not make sense to me at all.
>>
>> Best regards,
>> Krzysztof
>>
> Sorry if anything is unclear. I would try to explain the change from the driver's point of view.

I want "logical" point of view, not driver's.

> 
> According to i.MX7D Reference Manual, the watchdog controller allows, via the WDOGx_WCR register, control over whether the watchdog continues or suspends in:
> 
> a)WAIT power state (bit 7: WDW)
> b)STOP and DOSE power state(bit 0: WDZST).
> 
> The current driver implementation provides a Device Tree binding `fsl,suspend-in-wait` for configuring case (a) and forces the watchdog to be suspended in case (b).
> 
> My patch adds the ability to configure case (b) as well.

Just because you want to do something in the driver is not yet
justification we want it in DT. Why can't you enable it always? Why this
is system load (!!!) dependent?


Best regards,
Krzysztof