Re: [PATCH v5 5/5] watchdog: w83627hf_wdt: Add minute mode counting

["Paul Louvel" <paul.louvel@bootlin.com>]

Hi Guenter,

On Tue Oct 27, 2026 at 8:35 PM CET, Guenter Roeck wrote:
>> @@ -504,7 +523,7 @@ static int wdt_probe(struct platform_device *pdev)
>>   	wdd->ops = &wdt_ops;
>>   	wdd->timeout = WATCHDOG_TIMEOUT;
>>   	wdd->min_timeout = 1;
>> -	wdd->max_timeout = 255;
>> +	wdd->max_timeout = WATCHDOG_MAX_TIMEOUT;
>>   
>>   	switch (chip) {
>>   	case nct6102:
>> @@ -526,10 +545,27 @@ static int wdt_probe(struct platform_device *pdev)
>>   	watchdog_set_nowayout(wdd, nowayout);
>>   	watchdog_stop_on_reboot(wdd);
>>   
>> +	wdt_set_timeout(wdd, wdd->timeout);
>> +
>>   	ret = w83627hf_init(wdd, chip);
>>   	if (ret)
>>   		return dev_err_probe(dev, ret, "failed to initialize watchdog\n");
>>   
>> +	if (data->early_timer_val) {
>> +		if (early_disable) {
>
> The purpose of this flag is specifically to disable the watchdog
> "at module load time or kernel boot time". Moving it into the probe
> function would delay that and is therefore undesirable.
>
> On top of that, the flag is pointless in the probe function since
> the watchdog core takes care of pinging the watchdog until
> userspace kicks in if WDOG_HW_RUNNING is set.
>
> Also, any change made in that regard should be independent of
> introducing minute mode support.
>
> Thanks,
> Guenter
>

So, the w83627hf_init() function should be called inside wdt_init(). Then, this
is a problem introduced in patch #2.

Thanks.
Best regards,
-- 
Paul Louvel, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com