Re: [PATCH v5 06/10] drm/i915/hotplug: add helpers to track HPDs can generate PME

["Govindapillai, Vinod" <vinod.govindapillai@intel.com>]

On Fri, 2026-09-11 at 16:00 +0300, Jani Nikula wrote:
> On Fri, 11 Sep 2026, Vinod Govindapillai
> <vinod.govindapillai@intel.com> wrote:
> > As NVL+ devices can generate PME from HPDs, introduce a flag to
> > keep track of such a capability and helper functions to access
> > this flag. The plan is, this flag will be set to true from pm
> > runtime
> > suspend if PME is enabled and this flag will cleared on early
> > runtime
> > resume unconditionally. This is a place holder for the flag and
> > later
> > on this flag will be used for conditional IRQ resets and to decide
> > whether to do HPD polling.
> 
> What's the rationale for duplicating the state in a flag? Why can't
> you
> use intel_display_rpm_pme_enabled() and have a single point of truth?
> This would dodge the extra error path clearing too.
> 
> I don't understand.
> 
> BR,
> Jani.
> 

Ah..! Yes.. you are right. No rationale :( I kind of overlooked that
provision and was fixated on the steps I prepared before and fitting my
whole logic into that :(

Thanks. This is pretty clean. Will update. 
Patches 6, 7,and 8 are indeed redundant and will remove those and use
the rpm_pme_enabled directly in avoiding the HPD resets and HPD polling

+       bool reset_hpd = !intel_display_rpm_pme_enabled(display);

and

+       if (intel_display_rpm_pme_enabled(display)) {
+               drm_dbg_kms(display->drm, "PME wake capable device,
skipping HPD polling.\n");
+               return;
+       }


Before send a new revision, hope there are no other comments. FYI
@Jouni.

BR
Vinod

> > 
> > Bspec: 52979, 52980, 68857, 68867, 68970
> > Assisted-by: GitHub_Copilot:claude-opus-5
> > Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>