Re: [PATCH v5 10/10] drm/i915/hotplug: avoid HPD polling if the device is PME capable

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The patch fails to disable HPD polling for `xe` devices in D3cold because it relies on a stateful flag that is bypassed in the `xe` D3cold suspend path.
--

commit d64c6fbf549089ab8b3ecb3a04becd1d5a8da552
Author: Vinod Govindapillai <vinod.govindapillai@intel.com>

drm/i915/hotplug: avoid HPD polling if the device is PME capable

This commit attempts to optimize power usage by disabling HPD polling on
devices capable of generating PME wake events. It introduces a check in
intel_hpd_poll_enable() to skip polling if the device supports PME.

> diff --git a/drivers/gpu/drm/i915/display/intel_hotplug.c b/drivers/gpu/drm/i915/display/intel_hotplug.c
> index b24c9f360e241..8b7b9f47f049a 100644
> --- a/drivers/gpu/drm/i915/display/intel_hotplug.c
> +++ b/drivers/gpu/drm/i915/display/intel_hotplug.c
> @@ -877,6 +877,11 @@ void intel_hpd_poll_enable(struct intel_display *display)
>  	if (!HAS_DISPLAY(display) || !intel_display_device_enabled(display))
>  		return;
>  
> +	if (intel_hpd_can_generate_pme(display)) {
> +		drm_dbg_kms(display->drm, "PME wake capable device, skipping HPD polling.\n");
> +		return;
> +	}
> +

[Severity: Medium]
Will this check actually prevent HPD polling for xe devices entering D3cold?

When an xe device enters runtime suspend, xe_display_pm_runtime_suspend()
bypasses intel_display_driver_pm_runtime_suspend() if D3cold is allowed:

drivers/gpu/drm/xe/display/xe_display.c:xe_display_pm_runtime_suspend() {
	...
	if (xe->d3cold.allowed) {
		xe_display_enable_d3cold(xe);
		return;
	}

	intel_display_driver_pm_runtime_suspend(display);
	...
}

Because this bypasses intel_display_driver_pm_runtime_suspend(), the
display->hotplug.can_generate_pme flag is never populated for the D3cold path.

When xe_display_enable_d3cold() subsequently calls intel_hpd_poll_enable(),
intel_hpd_can_generate_pme() will return false, and HPD polling will remain
enabled, which appears to defeat the optimization this patch aims to achieve.

Does this flag need to be populated in the D3cold path as well?

>  	WRITE_ONCE(display->hotplug.poll_enabled, true);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260911080448.778316-1-vinod.govindapillai@intel.com?part=10