Am 23.11.2015 um 08:26 schrieb Guenter Roeck:
> On 11/22/2015 11:11 PM, Oleksij Rempel wrote:
>> Hi,
>>
>> thank you for review!
>>
>> Am 05.11.2015 um 21:43 schrieb Rob Herring:
>>> On Thu, Nov 05, 2015 at 10:06:56AM +0100, Oleksij Rempel wrote:
>>>> Add WD support for Alphascale asm9260 SoC. This driver
>>>> provide support for different function modes:
>>>> - HW mode to trigger SoC reset on timeout
>>>> - SW mode do soft reset if needed
>>>> - DEBUG mode
>>>>
>>>> Optional support for stopping watchdog. If reset binding are not
>>>> provided
>>>> this driver will work in nowayout mode.
>>>>
>>>> Signed-off-by: Oleksij Rempel <linux-YEK0n+YFykbzxQdaRaTXBw@public.gmane.org>
>>>> ---
>>>>   .../bindings/watchdog/alphascale-asm9260.txt       |  39 ++
>>>
>>> It is preferred that bindings are a separate patch.
>>>
>>>>   drivers/watchdog/Kconfig                           |   9 +
>>>>   drivers/watchdog/Makefile                          |   1 +
>>>>   drivers/watchdog/asm9260_wdt.c                     | 415
>>>> +++++++++++++++++++++
>>>>   4 files changed, 464 insertions(+)
>>>>   create mode 100644
>>>> Documentation/devicetree/bindings/watchdog/alphascale-asm9260.txt
>>>>   create mode 100644 drivers/watchdog/asm9260_wdt.c
>>>>
>>>> diff --git
>>>> a/Documentation/devicetree/bindings/watchdog/alphascale-asm9260.txt
>>>> b/Documentation/devicetree/bindings/watchdog/alphascale-asm9260.txt
>>>> new file mode 100644
>>>> index 0000000..6e54d1f
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/watchdog/alphascale-asm9260.txt
>>>> @@ -0,0 +1,39 @@
>>>> +Alphascale asm9260 Watchdog timer
>>>> +
>>>> +Required properties:
>>>> +
>>>> +- compatible : should be "alphascale,asm9260-wdt".
>>>> +- reg : Specifies base physical address and size of the registers.
>>>> +- clocks : the clocks feeding the watchdog timer. See
>>>> clock-bindings.txt
>>>> +- clock-names : should be set to
>>>> +    "mod" - source for tick counter.
>>>> +    "ahb" - ahb gate.
>>>> +
>>>> +Optional properties:
>>>> +- resets : phandle pointing to the system reset controller with
>>>> correct
>>>> +    reset line index for watchdog controller reset. This propertie is
>>>
>>> s/propertie/property/
>>>
>>>> +    required if you need to disable "nowayout" and it neened only with
>>>> +    CONFIG_WATCHDOG_NOWAYOUT=n.
>>>
>>> The DT cannot depend on certain kernel configs.
>>>
>>>> +    Without reseting this WD controller, it is not possible to stop
>>>
>>> s/reseting/resetting/
>>>
>>>> +    counter.
>>>> +- reset-names : should be set to "wdt_rst" if "resets" is used.
>>>> +- timeout-sec : shall contain the default watchdog timeout in seconds,
>>>> +    if unset, the default timeout is 30 seconds.
>>>> +- alphascale,mode : tree modes are supported
>>>
>>> s/tree/three/
>>>
>>>> +    "hw" - hw reset (defaul).
>>>
>>> s/defaul/default/
>>>
>>>> +    "sw" - sw reset.
>>>> +    "debug" - no action is taken.
>>>
>>> Seems like this should be a generic wdog property.
>>>
>>
>> what do you mean?
>> Using "mode" instead of "alphascale,mode" and implement it in WD
>> framework?
>>
> 
> You should still read the property in the driver, even it if is made
> generic.
> The watchdog framework can't really use the information, most drivers don't
> support selecting the mode. We can consider moving it into the framework
> if/when it benefits other drivers.
> 
> On a side note, I would suggest "software" and "hardware" instead of sw
> / hw,
> but that is just a personal preference.

Ok, then i will leave it for now as is. If "mode" will be introduced, it
can brake compatibility. Using "alphascale,mode" feels save :)

-- 
Regards,
Oleksij