Re: [PATCH] soc: qcom: pd_mapper: fix ADSP PD maps

[Dmitry Baryshkov]

On Fri, Sep 20, 2024 at 07:00:11AM GMT, Chris Lew wrote:
> 
> 
> On 9/20/2024 2:02 AM, Johan Hovold wrote:
> > On Fri, Sep 20, 2024 at 11:49:46AM +0300, Dmitry Baryshkov wrote:
> > > On Fri, Sep 20, 2024 at 10:21:03AM GMT, Johan Hovold wrote:
> > > > On Wed, Sep 18, 2024 at 04:02:39PM +0300, Dmitry Baryshkov wrote:
> > > > > On X1E8 devices root ADSP domain should have tms/pdr_enabled registered.
> > > > > Change the PDM domain data that is used for X1E80100 ADSP.
> > > > 
> > > > Please expand the commit message so that it explains why this is
> > > > needed and not just describes what the patch does.
> > > 
> > > Unfortunately in this case I have no idea. It marks the domain as
> > > restartable (?), this is what json files for CRD and T14s do. Maybe
> > > Chris can comment more.
> > 
> > Chris, could you help sort out if and why this change is needed?
> > 
> > 	https://lore.kernel.org/all/20240918-x1e-fix-pdm-pdr-v1-1-cefc79bb33d1@xxxxxxxxxx/	
> > 
> 
> I don't think this change would help with the issue reported by Johan. From
> a quick glance, I couldn't find where exactly the restartable attribute is
> used, but this type of change would only matter when the ChargerPD is
> started or restarted.

This raises a question: should we care at all about the pdr_enabled? Is
it fine to drop it fromm all PD maps?

> 
> The PMIC_GLINK channel probing in rpmsg is dependent on ChargerPD starting,
> so we know ChargerPD can start with or without this change.
> 
> I can give this change a try next week to help give a better analysis.
> 
> > > > What is the expected impact of this and is there any chance that this is
> > > > related to some of the in-kernel pd-mapper regression I've reported
> > > > (e.g. audio not being registered and failing with a PDR error)?
> > > > 
> > > > 	https://lore.kernel.org/all/ZthVTC8dt1kSdjMb@xxxxxxxxxxxxxxxxxxxx/
> > > 
> > > Still debugging this, sidetracked by OSS / LPC.
> > 
> > Johan

-- 
With best wishes
Dmitry