Re: [PATCH v5 4/6] OMAP4: hwmod data: Add watchdog timer

[Kevin Hilman <khilman@deeprootsystems.com>]

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

> Hi Charu,
>
> On 9/17/2010 11:23 AM, Varadarajan, Charulatha wrote:
>>
>>
>>> -----Original Message-----
>>> From: Cousson, Benoit
>>> Sent: Thursday, September 16, 2010 2:49 AM
>>> To: Varadarajan, Charulatha
>>> Cc: tony@atomide.com; wim@iguana.be; linux-watchdog@vger.kernel.org;
>>> khilman@deeprootsystems.com; paul@pwsan.com; Nayak, Rajendra; Basak,
>>> Partha; linux-omap@vger.kernel.org
>>> Subject: Re: [PATCH v5 4/6] OMAP4: hwmod data: Add watchdog timer
>>>
>>> Hi Charu,
>>>
>>> On 9/15/2010 7:26 AM, Varadarajan, Charulatha wrote:
>>>> From: Benoit Cousson<b-cousson@ti.com>
>>>>
>>>> Add watchdog timer hwmod data for OMAP4 chip
>>>>
>>>> Signed-off-by: Benoit Cousson<b-cousson@ti.com>
>>>> Signed-off-by: Charulatha V<charu@ti.com>
>>>> ---
>> <<snip>>
>>
>>>> @@ -472,6 +538,8 @@ static __initdata struct omap_hwmod
>>> *omap44xx_hwmods[] = {
>>>>
>>>>    	/* mpu class */
>>>>    	&omap44xx_mpu_hwmod,
>>>> +	/* wd_timer class */
>>>> +	&omap44xx_wd_timer2_hwmod,
>>>>    	NULL,
>>>>    };
>>>
>>> What about the wd_timer3? It is in the audio backend, and I'm not sure
>>> who will use it, but it does exist.
>>
>> As you mentioned, wd_timer3 in OMAP4 is an IVA3 watchdog timer.
>>
>> Also the current watchdog driver (omap_wdt.c) does not support more than one watchdog timer. To add this support we need to first redesign the driver code and then add wd_timer3 to the database.
>>
>> This patch series is intended to make watchdog driver to use omap_hwmod APIs and PM runtime APIs. Driver modification to support more than one watchdog timers can be taken up as a separate activity later.
>
> No necessarily, even if you do not manage this watchdog, we do need to
> expose its hwmod in order to allow dspbridge/syslink to potentially
> enable or disable it.

We also should expose it in the hwmod db so that it is initialized and
properly reset on init.  

Later on bridge/syslink can initialize and use as desired.

Kevin

> Maybe, it should be up to syslink people to do that then.
> In that case, please add comment in the DB to explain that we do have
> other instances but they are not managed yet.
>
> Thanks,
> Benoit