Re: [PATCH] clk: qcom: camcc-sc8280xp: Remove always-on GDSC hard-coding

[Satya Priya Kakitapalli (Temp)]

On 7/27/2024 2:01 AM, Bryan O'Donoghue wrote:
> On 26/07/2024 08:01, Satya Priya Kakitapalli (Temp) wrote:
> > On 7/23/2024 2:59 PM, Bryan O'Donoghue wrote:
> > > On 22/07/2024 09:57, Satya Priya Kakitapalli (Temp) wrote:
> > > > > I have no idea. Why does it matter ?
> > > >
> > > > This clock expected to be kept always ON, as per design, or else 
> > > > the GDSC transition form ON to OFF (vice versa) wont work.
> > >
> > > Yes, parking to XO per this patch works for me. So I guess its 
> > > already on and is left in that state by the park.
> > >
> > > > Want to know the clock status after bootup, to understand if the 
> > > > clock got turned off during the late init. May I know exactly what 
> > > > you have tested? Did you test the camera usecases as well?
> > >
> > > Of course.
> > >
> > > The camera works on x13s with this patch. That's what I mean by tested.
> >
> > It might be working in your case, but it is not the HW design 
> > recommended way to do. The same should not be propagated to other 
> > target's camcc drivers, as I already observed it is not working on 
> > SM8150.
>
> I don't think the argument here really stands up.
>
> We've established that the GDSC clock and PDs will remain on when the 
> clock gets parked right ?
>
> Am I missing something obvious here ?

Yes, just parking the RCG at XO, does not ensure the branch clock, i.e, 
the 'cam_cc_gdsc_clk' as always ON. When I compiled the camcc-sm8150 
driver statically, I see that the clock is getting disabled in late_init 
if it is modelled.


Can you confirm if the camcc-sc8280xp driver is compiled statically or 
as a module at your end?


> ---
> bod