[PATCH v5 0/4] add meson secure watchdog driver

[PATCH v5 0/4] add meson secure watchdog driver

[Xingyu Chen]

The watchdog controller on the Meson-A/C series SoCs is moved to secure world,
We have to call SMC instruction to trap the ATF for watchdog operation. These
operations are different from previous SoCs, so we introduce a new watchdog
driver to support this kind of SoCs.

Changes since v4 at [3]:
- add watchdog node in dts, and introduce a new optional property "timeout-sec"
- add dt-binding for meson secure watchdog
- instantiate wdt device through dts node instead of platform_device_register_simple()

Changes since v3 at [2]:
- add SM_A1_ prefix for WATCHDOG_OPS
- remove phandle to secure-monitor node
- remove watchdog node from dts, and register wdt device by platform_device_register_simple()
- remove dt-binding for meson secure watchdog
- use the msec as unit of timeout parameter which is passed to fw side

Changes since v2 at [1]:
- remove useless dependency in Kconfig
- return zero when getting left time value fails

Changes since v1 at [0]:
- add a new dependency in Kconfig
- simplify/add the return operation
- remove useless ping operation when setting the timeout
- fix some return values
- fix the license statement

[0]:https://lore.kernel.org/linux-amlogic/1570874721-36077-1-git-send-email-xingyu.chen@amlogic.com
[1]:https://lore.kernel.org/linux-amlogic/1571387622-35132-1-git-send-email-xingyu.chen@amlogic.com
[2]:https://lore.kernel.org/linux-amlogic/1571983984-11771-1-git-send-email-xingyu.chen@amlogic.com
[3]:https://lore.kernel.org/linux-amlogic/1574685218-31164-1-git-send-email-xingyu.chen@amlogic.com

Xingyu Chen (4):
  firmware: meson_sm: add new SMC ID support for accessing secure
    watchdog
  dt-bindings: watchdog: add new binding for meson secure watchdog
  watchdog: add meson secure watchdog driver
  arm64: dts: a1: add secure watchdog controller

 .../bindings/watchdog/amlogic,meson-sec-wdt.yaml   |  35 ++++
 arch/arm64/boot/dts/amlogic/meson-a1.dtsi          |   6 +
 drivers/firmware/meson/meson_sm.c                  |   1 +
 drivers/watchdog/Kconfig                           |  16 ++
 drivers/watchdog/Makefile                          |   1 +
 drivers/watchdog/meson_sec_wdt.c                   | 188 +++++++++++++++++++++
 include/linux/firmware/meson/meson_sm.h            |   1 +
 7 files changed, 248 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/watchdog/amlogic,meson-sec-wdt.yaml
 create mode 100644 drivers/watchdog/meson_sec_wdt.c

-- 
2.7.4

[PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Xingyu Chen]

The binding targets the Meson-A/C series compatible SoCs, in which the
watchdog registers are in secure world.

Signed-off-by: Xingyu Chen <xingyu.chen@amlogic.com>
---
 .../bindings/watchdog/amlogic,meson-sec-wdt.yaml   | 35 ++++++++++++++++++++++
 1 file changed, 35 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/watchdog/amlogic,meson-sec-wdt.yaml

diff --git a/Documentation/devicetree/bindings/watchdog/amlogic,meson-sec-wdt.yaml b/Documentation/devicetree/bindings/watchdog/amlogic,meson-sec-wdt.yaml
new file mode 100644
index 00000000..c0ff2ec
--- /dev/null
+++ b/Documentation/devicetree/bindings/watchdog/amlogic,meson-sec-wdt.yaml
@@ -0,0 +1,35 @@
+# SPDX-License-Identifier: (GPL-2.0+ OR MIT)
+# Copyright (c) 2019 Amlogic, Inc
+%YAML 1.2
+---
+$id: "http://devicetree.org/schemas/power/amlogic,meson-sec-wdt.yaml#"
+$schema: "http://devicetree.org/meta-schemas/core.yaml#"
+
+title: Amlogic Meson Secure Watchdog Timer
+
+maintainers:
+  - Xingyu Chen <xingyu.chen@amlogic.com>
+
+description: |+
+  Secure Watchdog Timer used in Meson-A/C series Compatible SoCs
+
+properties:
+  compatible:
+    enum:
+      - amlogic,meson-sec-wdt
+
+  timeout-sec:
+    minimum: 1
+    maximum: 65
+    description:
+      Watchdog timeout values (in seconds)
+
+required:
+  - compatible
+
+examples:
+  - |
+    watchdog {
+          compatible = "amlogic,meson-sec-wdt";
+          timeout-sec = <60>;
+    };
-- 
2.7.4

[PATCH v5 4/4] arm64: dts: a1: add secure watchdog controller

[Xingyu Chen]

Enable secure watchdog controller for Meson-A1 SoC

Signed-off-by: Xingyu Chen <xingyu.chen@amlogic.com>
---
 arch/arm64/boot/dts/amlogic/meson-a1.dtsi | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/arch/arm64/boot/dts/amlogic/meson-a1.dtsi b/arch/arm64/boot/dts/amlogic/meson-a1.dtsi
index 4dec518..1e354b7 100644
--- a/arch/arm64/boot/dts/amlogic/meson-a1.dtsi
+++ b/arch/arm64/boot/dts/amlogic/meson-a1.dtsi
@@ -118,6 +118,12 @@
 				clock-names = "xtal", "pclk", "baud";
 				status = "disabled";
 			};
+
+			watchdog {
+				compatible = "amlogic,meson-sec-wdt";
+				timeout-sec = <60>;
+				status = "okay";
+			};
 		};
 
 		gic: interrupt-controller@ff901000 {
-- 
2.7.4

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Martin Blumenstingl]

Hi Xingyu and Rob,

On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen <xingyu.chen@amlogic.com> wrote:
[...]
> +examples:
> +  - |
> +    watchdog {
> +          compatible = "amlogic,meson-sec-wdt";
> +          timeout-sec = <60>;
> +    };
in v3 of this patch Rob commented that there shouldn't be an OF node
if there are no additional properties
with timeout-sec there's now an additional property so my
understanding is that it's fine to have an OF node

what I don't understand yet is where this node should be placed.
is it supposed to be a child node of the secure monitor node (for
which we already have a binding here:
Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
where else would we place it inside the .dts?


Martin


[0] https://patchwork.kernel.org/patch/11211399/

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Xingyu Chen]

Hi, Martin

Sorry for the late reply.

On 2019/12/13 4:05, Martin Blumenstingl wrote:
> Hi Xingyu and Rob,
>
> On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen <xingyu.chen@amlogic.com> wrote:
> [...]
>> +examples:
>> +  - |
>> +    watchdog {
>> +          compatible = "amlogic,meson-sec-wdt";
>> +          timeout-sec = <60>;
>> +    };
> in v3 of this patch Rob commented that there shouldn't be an OF node
> if there are no additional properties
> with timeout-sec there's now an additional property so my
> understanding is that it's fine to have an OF node
Your understanding is correct.
>
> what I don't understand yet is where this node should be placed.
> is it supposed to be a child node of the secure monitor node (for
> which we already have a binding here:
> Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
> where else would we place it inside the .dts?
IMO,  Although the watchdog node need to reference the meson_sm node, 
there is no
bus-like dependencies between the devices which the two nodes 
corresponding to.
so i think that the watchdog node as child node of meson_sm maybe not 
appropriate.
>
>
> Martin
>
>
> [0] https://patchwork.kernel.org/patch/11211399/
>
> .
>

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Guenter Roeck]

On 12/15/19 10:03 PM, Xingyu Chen wrote:
> Hi, Martin
> 
> Sorry for the late reply.
> 
> On 2019/12/13 4:05, Martin Blumenstingl wrote:
>> Hi Xingyu and Rob,
>>
>> On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen <xingyu.chen@amlogic.com> wrote:
>> [...]
>>> +examples:
>>> +  - |
>>> +    watchdog {
>>> +          compatible = "amlogic,meson-sec-wdt";
>>> +          timeout-sec = <60>;
>>> +    };
>> in v3 of this patch Rob commented that there shouldn't be an OF node
>> if there are no additional properties
>> with timeout-sec there's now an additional property so my
>> understanding is that it's fine to have an OF node
> Your understanding is correct.
>>
>> what I don't understand yet is where this node should be placed.
>> is it supposed to be a child node of the secure monitor node (for
>> which we already have a binding here:
>> Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
>> where else would we place it inside the .dts?
> IMO,  Although the watchdog node need to reference the meson_sm node, there is no
> bus-like dependencies between the devices which the two nodes corresponding to.
> so i think that the watchdog node as child node of meson_sm maybe not appropriate.

The watchdog driver needs the meson SM's dt node, and it depends on the existence
of that node. That seems enough of a relationship to warrant having it as child note.

Guenter

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Xingyu Chen]

Hi, Guenter Martin

On 2019/12/16 21:30, Guenter Roeck wrote:
> On 12/15/19 10:03 PM, Xingyu Chen wrote:
>> Hi, Martin
>>
>> Sorry for the late reply.
>>
>> On 2019/12/13 4:05, Martin Blumenstingl wrote:
>>> Hi Xingyu and Rob,
>>>
>>> On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen 
>>> <xingyu.chen@amlogic.com> wrote:
>>> [...]
>>>> +examples:
>>>> +  - |
>>>> +    watchdog {
>>>> +          compatible = "amlogic,meson-sec-wdt";
>>>> +          timeout-sec = <60>;
>>>> +    };
>>> in v3 of this patch Rob commented that there shouldn't be an OF node
>>> if there are no additional properties
>>> with timeout-sec there's now an additional property so my
>>> understanding is that it's fine to have an OF node
>> Your understanding is correct.
>>>
>>> what I don't understand yet is where this node should be placed.
>>> is it supposed to be a child node of the secure monitor node (for
>>> which we already have a binding here:
>>> Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
>>> where else would we place it inside the .dts?
>> IMO,  Although the watchdog node need to reference the meson_sm node, 
>> there is no
>> bus-like dependencies between the devices which the two nodes 
>> corresponding to.
>> so i think that the watchdog node as child node of meson_sm maybe not 
>> appropriate.
>
> The watchdog driver needs the meson SM's dt node, and it depends on 
> the existence
> of that node. That seems enough of a relationship to warrant having it 
> as child note.
Thanks for your reply, if i take the wdt node as child of secure monitor 
(sm), how should
i register or find the wdt device ?

I only think of the following three methods :
1). update the sm driver，and scan&register wdt device when the sm driver 
probes(It is like i2c), but there
are too many changes involved.

2). add "simple-bus" key string to compatible of sm node, and it will 
make the child node is registered as
platform device, but it seems that the key string is not match current 
scene.

secure-monitor {
     compatible = "amlogic,meson-gxbb-sm",  "simple-bus";

     watchdog {
         compatible = "amlogic,meson-sec-wdt";
         timeout-sec = <60>;
     }
}

3).  don't register device, and find directly the watchdog node by using 
the of_* API in watchdog
driver (Eg: linux-4.x/drivers/tee/optee/core.c)

secure-monitor {
     compatible = "amlogic,meson-gxbb-sm";

     watchdog {
         compatible = "amlogic,meson-sec-wdt";
         timeout-sec = <60>;
     }
}

The method 3 looks better for me, do you have a better suggestion ? Thanks

BR
>
> Guenter
>
> .
>

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Rob Herring]

On Wed, Dec 18, 2019 at 03:37:49PM +0800, Xingyu Chen wrote:
> Hi, Guenter Martin
> 
> On 2019/12/16 21:30, Guenter Roeck wrote:
> > On 12/15/19 10:03 PM, Xingyu Chen wrote:
> > > Hi, Martin
> > > 
> > > Sorry for the late reply.
> > > 
> > > On 2019/12/13 4:05, Martin Blumenstingl wrote:
> > > > Hi Xingyu and Rob,
> > > > 
> > > > On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen
> > > > <xingyu.chen@amlogic.com> wrote:
> > > > [...]
> > > > > +examples:
> > > > > +  - |
> > > > > +    watchdog {
> > > > > +          compatible = "amlogic,meson-sec-wdt";
> > > > > +          timeout-sec = <60>;
> > > > > +    };
> > > > in v3 of this patch Rob commented that there shouldn't be an OF node
> > > > if there are no additional properties
> > > > with timeout-sec there's now an additional property so my
> > > > understanding is that it's fine to have an OF node
> > > Your understanding is correct.
> > > > 
> > > > what I don't understand yet is where this node should be placed.
> > > > is it supposed to be a child node of the secure monitor node (for
> > > > which we already have a binding here:
> > > > Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
> > > > where else would we place it inside the .dts?
> > > IMO,  Although the watchdog node need to reference the meson_sm
> > > node, there is no
> > > bus-like dependencies between the devices which the two nodes
> > > corresponding to.
> > > so i think that the watchdog node as child node of meson_sm maybe
> > > not appropriate.
> > 
> > The watchdog driver needs the meson SM's dt node, and it depends on the
> > existence
> > of that node. That seems enough of a relationship to warrant having it
> > as child note.
> Thanks for your reply, if i take the wdt node as child of secure monitor
> (sm), how should
> i register or find the wdt device ?
> 
> I only think of the following three methods :
> 1). update the sm driver，and scan&register wdt device when the sm driver
> probes(It is like i2c), but there
> are too many changes involved.

Just add of_platform_default_populate() call and clean-up calls. That's 
not what I'd call 'too many changes'.

 
> 2). add "simple-bus" key string to compatible of sm node, and it will make
> the child node is registered as
> platform device, but it seems that the key string is not match current
> scene.

You previously said it's not a bus...

> 
> secure-monitor {
>     compatible = "amlogic,meson-gxbb-sm",  "simple-bus";
> 
>     watchdog {
>         compatible = "amlogic,meson-sec-wdt";
>         timeout-sec = <60>;
>     }
> }
> 
> 3).  don't register device, and find directly the watchdog node by using the
> of_* API in watchdog
> driver (Eg: linux-4.x/drivers/tee/optee/core.c)
> 
> secure-monitor {
>     compatible = "amlogic,meson-gxbb-sm";
> 
>     watchdog {
>         compatible = "amlogic,meson-sec-wdt";
>         timeout-sec = <60>;
>     }
> }
> 
> The method 3 looks better for me, do you have a better suggestion ? Thanks
> 
> BR
> > 
> > Guenter
> > 
> > .
> > 

Re: [PATCH v5 2/4] dt-bindings: watchdog: add new binding for meson secure watchdog

[Xingyu Chen]

Hi, Rob

On 2019/12/19 4:34, Rob Herring wrote:
> On Wed, Dec 18, 2019 at 03:37:49PM +0800, Xingyu Chen wrote:
>> Hi, Guenter Martin
>>
>> On 2019/12/16 21:30, Guenter Roeck wrote:
>>> On 12/15/19 10:03 PM, Xingyu Chen wrote:
>>>> Hi, Martin
>>>>
>>>> Sorry for the late reply.
>>>>
>>>> On 2019/12/13 4:05, Martin Blumenstingl wrote:
>>>>> Hi Xingyu and Rob,
>>>>>
>>>>> On Thu, Dec 12, 2019 at 1:20 PM Xingyu Chen
>>>>> <xingyu.chen@amlogic.com> wrote:
>>>>> [...]
>>>>>> +examples:
>>>>>> +  - |
>>>>>> +    watchdog {
>>>>>> +          compatible = "amlogic,meson-sec-wdt";
>>>>>> +          timeout-sec = <60>;
>>>>>> +    };
>>>>> in v3 of this patch Rob commented that there shouldn't be an OF node
>>>>> if there are no additional properties
>>>>> with timeout-sec there's now an additional property so my
>>>>> understanding is that it's fine to have an OF node
>>>> Your understanding is correct.
>>>>> what I don't understand yet is where this node should be placed.
>>>>> is it supposed to be a child node of the secure monitor node (for
>>>>> which we already have a binding here:
>>>>> Documentation/devicetree/bindings/firmware/meson/meson_sm.txt) or
>>>>> where else would we place it inside the .dts?
>>>> IMO,  Although the watchdog node need to reference the meson_sm
>>>> node, there is no
>>>> bus-like dependencies between the devices which the two nodes
>>>> corresponding to.
>>>> so i think that the watchdog node as child node of meson_sm maybe
>>>> not appropriate.
>>> The watchdog driver needs the meson SM's dt node, and it depends on the
>>> existence
>>> of that node. That seems enough of a relationship to warrant having it
>>> as child note.
>> Thanks for your reply, if i take the wdt node as child of secure monitor
>> (sm), how should
>> i register or find the wdt device ?
>>
>> I only think of the following three methods :
>> 1). update the sm driver，and scan&register wdt device when the sm driver
>> probes(It is like i2c), but there
>> are too many changes involved.
> Just add of_platform_default_populate() call and clean-up calls. That's
> not what I'd call 'too many changes'.
Thanks for your guidance.
>
>   
>> 2). add "simple-bus" key string to compatible of sm node, and it will make
>> the child node is registered as
>> platform device, but it seems that the key string is not match current
>> scene.
> You previously said it's not a bus...
>
>> secure-monitor {
>>      compatible = "amlogic,meson-gxbb-sm",  "simple-bus";
>>
>>      watchdog {
>>          compatible = "amlogic,meson-sec-wdt";
>>          timeout-sec = <60>;
>>      }
>> }
>>
>> 3).  don't register device, and find directly the watchdog node by using the
>> of_* API in watchdog
>> driver (Eg: linux-4.x/drivers/tee/optee/core.c)
>>
>> secure-monitor {
>>      compatible = "amlogic,meson-gxbb-sm";
>>
>>      watchdog {
>>          compatible = "amlogic,meson-sec-wdt";
>>          timeout-sec = <60>;
>>      }
>> }
>>
>> The method 3 looks better for me, do you have a better suggestion ? Thanks
>>
>> BR
>>> Guenter
>>>
>>> .
>>>
> .
>