[QUESTION] How to set static/dynamic dependency between clock domains

[Shilimkar, Santosh]

On Fri, Dec 16, 2011 at 9:16 PM, Ming Lei <tom.leiming at gmail.com> wrote:
> On Fri, Dec 16, 2011 at 11:13 PM, Shilimkar, Santosh
> <santosh.shilimkar at ti.com> wrote:
>> On Fri, Dec 16, 2011 at 7:20 PM, Ming Lei <tom.leiming at gmail.com> wrote:
>>> Hi,
>>>
>>> On Fri, Dec 16, 2011 at 9:24 PM, Ming Lei <tom.leiming at gmail.com> wrote:
>>>
>>>> [1], add static dependency
>>>
>>> Sorry for the mess, see attachment for the change.
>>>
>> Your patch is setting static dependency between ISS and l3_2.
>> It's not dynamic dep. Dynamic dep is managed through hardware and they
>
> Yes, I see now,  L3_2_DYNDEP of CM_CAM_DYNAMICDEP is read only.
>
>> are not configurable. But with static dep. you can over-ride that behavior
>> as you have done.
>>
>> This will impact power since l3_2 can't idle as long as iss clock
>> domain is active.
>>
>> I need to check internal code-base but I don't remember this
>> issue observed so far.
>
> In fact, I saw the issue on 3.2.0-rc5-next-20111216, and even
> no such issue on 3.2.0-rc5.  After some bisecting, I found below
> is the first commit on which the issue can be observed:
>
> commit 3c50729b3fa1cd8ca1f347e6caf1081204cf1a7c
> Author: Santosh Shilimkar <santosh.shilimkar at ti.com>
> Date:   Wed Jan 5 22:03:17 2011 +0530
>
>    ARM: OMAP4: PM: Initialise all the clockdomains to supported states
>
>    Initialise hardware supervised mode for all clockdomains if it's
>    supported. Initiate sleep transition for other clockdomains,
>    if they are not being used.
>
Before this patch all clock-domains were  put into SW forcewakeup
and hence the clock domains were not idling. With this patch now, hw
will take control of the clock-domain idle based on the domain
activities.

Regards
Santosh