[PATCH] OMAP2PLUS: WDT: Fix: Disable WDT after reset during init

[Kevin Hilman]

"Cousson, Benoit" <b-cousson at ti.com> writes:

> On 9/30/2010 3:55 PM, Kevin Hilman wrote:
>> "Cousson, Benoit"<b-cousson at ti.com>  writes:
>>
>>> Hi Charu,
>>>
>>> On 9/30/2010 10:11 AM, Varadarajan, Charulatha wrote:
>>>> With OMAP2PLUS watchdog implemented in hwmod fw way, the
>>>> module is reset during init.
>>>
>>> In that case hwmod fw just highlighted the real behavior that was
>>> hidden so far by the X-loader.
>>>
>>> You should as well add a link to the email thread with Kevin that
>>> raised the issue.
>>>
>>>> After a watchdog timer module reset, the WDTs are enabled. The
>>>> default time for a system reset after a watchdog module reset
>>>> is ~10s as per the default value of the WDT registers. Hence
>>>> the system would be reset after 10s, if watchdog is not disabled
>>>> within 10s.
>>>>
>>>> This patch fixes the above issue by disabling the watchdog timer
>>>> after reset during initialization of devices.
>>>
>>> I'm still wondering as well what is the expected behavior of the
>>> watchdog in a real product. If it is started by default at boot time,
>>> this is probably for a good reason (or maybe not...).
>>>
>>> So, disabling it all the time is maybe not the best solution.
>>
>> I'm not sure what the other options are.  If you don't have a watchdog
>> driver, and the watchdog is armed, it will reboot the system.
>>
>> The approach in this patch is just to continue the behavior that all
>> bootloaders currently do, but make it explicit in the kernel.
>
> Yes, because we are not a building a product, and for us watchdog is a
> pain. But I'm not sure that a real product will disable that at all
> during the boot process.

Maybe not.  But wouldn't a product just ensure the real watchdog
driver is loaded?

Or more likely, in addition to the hundreds of other out-of-tree
patches, they would just remove $SUBJECT patch.  ;)

> I think that disabling it should be done only if the CONFIG_OMAP_WDT
> is not set.

That's not easy either, as the watchdog driver can be built as a module,
and may (or may not) be loaded some unknown time from the kernel boot,
resulting in a likely

Kevin

> But since I don't have a clue about a product can use that, it will be
> good to have such inputs to understand the usecase.