On 02/10/2022 18:29, Guenter Roeck wrote:
> On 9/30/22 03:35, Krzysztof Kozlowski wrote:
>> On 27/09/2022 13:02, Srinivas Neeli wrote:
>>> +
>>> +static void xwwdt_clk_disable_unprepare(void *data)
>>> +{
>>> + clk_disable_unprepare(data);
>>
>> If watchdog is stopped and then device unbound, don't you have double
>> disable? IOW, where is matching clk_enable?
>>
>
> After looking into the entire driver code, I see the problem:
> clk_disable() is called in the stop function. That is wrong;
> if the clock is disabled in the stop function, it needs to be
Yep. clk_disable and clk_enable are always paired. In probe they are
pared with remove() or reset-action. This one is in stop(), so it does
not have its pair.
Best regards,
Krzysztof
