Re: [CI] drm/{i915, xe}: move intel_opregion_suspend() to intel_display_driver_pm_suspend_late()

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Thu, Oct 08, 2026 at 05:00:04PM +0300, Jani Nikula wrote:
> On Thu, 08 Oct 2026, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote:
> > On Thu, Oct 08, 2026 at 02:47:35PM +0300, Jani Nikula wrote:
> >> Try to unify the suspend paths and behaviour by moving the
> >> intel_opregion_suspend() calls to
> >> intel_display_driver_pm_suspend_late().
> >> 
> >> This is a change in the suspend sequences. The idea behind adding this
> >> to intel_display_driver_pm_suspend_late() is primarily based on not
> >> having to pass the s2idle parameter to more functions than that.
> >> 
> >> This also changes behaviour for i915 hibernation, going for PCI_D3cold
> >> instead of PCI_D1 on hibernate.
> >
> > Pretty sure I pointed out before that this isn't true. PCI_D3cold is
> > what the current code also does.
> 
> Right, you did, but I failed to address it. Sorry.
> 
> I have tried to track down what acpi_target_system_state() actually
> returns during .freeze_late and .poweroff_late struct dev_pm_ops calls,
> and it's far from obvious.
> 
> I don't think you can reliably say acpi_target_system_state() ==
> ACPI_STATE_S4 on the paths that pass hibernation=true.

After staring at it a bit I think what I said is correct.

hibernation_snapshot() calls acpi_hibernation_begin() (assuming
/sys/power/disk==platform) prior to dpm_suspend(PMSG_FREEZE),
and hibernation_platform_enter() calls it again prior to
dpm_suspend_start(PMSG_HIBERNATE).

Also verified on an actual system, and both .freeze() and
.poweroff() see PCI_D3cold.

Looks like the thing that does change a bit is whether opregion
suspend happens at all during error handling and some pm_test
modes, because the _late() hooks may be skipped for those.

> 
> Are we good to go with the patch or do you want some changes?
> 
> 
> BR,
> Jani.
> 
> 
> >
> > I suppose it *might* be the case if one does something like 
> > 'echo reboot > /sys/power/disk', but I've not actually checked
> > what that sort of thing does to acpi_target_system_state()...
> >
> >> It's probably the correct thing to do
> >> anyway, I don't think it should matter all that much, but fingers
> >> crossed.
> >> 
> >> Reviewed-by: Imre Deak <imre.deak@intel.com>
> >> Signed-off-by: Jani Nikula <jani.nikula@intel.com>
> >> ---
> >>  drivers/gpu/drm/i915/display/intel_display_driver.c | 2 ++
> >>  drivers/gpu/drm/i915/i915_driver.c                  | 4 ----
> >>  drivers/gpu/drm/xe/display/xe_display.c             | 3 ---
> >>  3 files changed, 2 insertions(+), 7 deletions(-)
> >> 
> >> diff --git a/drivers/gpu/drm/i915/display/intel_display_driver.c b/drivers/gpu/drm/i915/display/intel_display_driver.c
> >> index 77009dca7d0d..abab457bce3a 100644
> >> --- a/drivers/gpu/drm/i915/display/intel_display_driver.c
> >> +++ b/drivers/gpu/drm/i915/display/intel_display_driver.c
> >> @@ -783,6 +783,8 @@ void intel_display_driver_pm_suspend_late(struct intel_display *display, bool s2
> >>  	if (!HAS_DISPLAY(display))
> >>  		return;
> >>  
> >> +	intel_opregion_suspend(display, s2idle ? PCI_D1 : PCI_D3cold);
> >> +
> >>  	intel_display_power_suspend_late(display, s2idle);
> >>  }
> >>  
> >> diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c
> >> index 9e047c27a153..293099fe6639 100644
> >> --- a/drivers/gpu/drm/i915/i915_driver.c
> >> +++ b/drivers/gpu/drm/i915/i915_driver.c
> >> @@ -1075,7 +1075,6 @@ static int i915_drm_suspend(struct drm_device *dev)
> >>  {
> >>  	struct drm_i915_private *dev_priv = to_i915(dev);
> >>  	struct intel_display *display = dev_priv->display;
> >> -	pci_power_t opregion_target_state;
> >>  
> >>  	disable_rpm_wakeref_asserts(&dev_priv->runtime_pm);
> >>  
> >> @@ -1089,9 +1088,6 @@ static int i915_drm_suspend(struct drm_device *dev)
> >>  
> >>  	i9xx_display_sr_save(display);
> >>  
> >> -	opregion_target_state = suspend_to_idle(dev_priv) ? PCI_D1 : PCI_D3cold;
> >> -	intel_opregion_suspend(display, opregion_target_state);
> >> -
> >>  	dev_priv->suspend_count++;
> >>  
> >>  	enable_rpm_wakeref_asserts(&dev_priv->runtime_pm);
> >> diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c
> >> index aab416485e3e..7d19f3c42e0e 100644
> >> --- a/drivers/gpu/drm/xe/display/xe_display.c
> >> +++ b/drivers/gpu/drm/xe/display/xe_display.c
> >> @@ -238,14 +238,11 @@ static bool suspend_to_idle(void)
> >>  void xe_display_pm_suspend(struct xe_device *xe)
> >>  {
> >>  	struct intel_display *display = xe->display;
> >> -	bool s2idle = suspend_to_idle();
> >>  
> >>  	if (!xe->info.probe_display)
> >>  		return;
> >>  
> >>  	intel_display_driver_pm_suspend(display);
> >> -
> >> -	intel_opregion_suspend(display, s2idle ? PCI_D1 : PCI_D3cold);
> >>  }
> >>  
> >>  void xe_display_pm_suspend_late(struct xe_device *xe)
> >> -- 
> >> 2.47.3
> 
> -- 
> Jani Nikula, Intel

-- 
Ville Syrjälä
Intel