| Message ID | 20240903223803.380711-5-rodrigo.vivi@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/xe/display: Clean-up PM sequences | expand |
On 2024-09-04 at 04:08:03 +0530, Rodrigo Vivi wrote: > When going to d3cold we surely don't need to flush commits, > nor handle fbdev. So, clean this up a bit. > > Also move the opregion call to make the function more symmetric. > > Cc: Imre Deak <imre.deak@intel.com> > Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> > --- > drivers/gpu/drm/xe/display/xe_display.c | 12 +++--------- > 1 file changed, 3 insertions(+), 9 deletions(-) > > diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c > index daf9b9baa88a..3fd3e74f1310 100644 > --- a/drivers/gpu/drm/xe/display/xe_display.c > +++ b/drivers/gpu/drm/xe/display/xe_display.c > @@ -310,9 +310,6 @@ static void xe_display_to_d3cold(struct xe_device *xe) > * properly. > */ > intel_power_domains_disable(xe); > - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_SUSPENDED, true); How do we make sure fbconsole doesn't use vram to blink the cursor ? Thanks, Anshuman. > - > - xe_display_flush_cleanup_work(xe); > > intel_dp_mst_suspend(xe); > > @@ -329,18 +326,15 @@ static void xe_display_from_d3cold(struct xe_device *xe) > > intel_dmc_resume(xe); > > - drm_mode_config_reset(&xe->drm); > - > intel_display_driver_init_hw(xe); > + > + intel_opregion_resume(display); > + > intel_hpd_init(xe); > > /* MST sideband requires HPD interrupts enabled */ > intel_dp_mst_resume(xe); > > - intel_opregion_resume(display); > - > - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_RUNNING, false); > - > intel_power_domains_enable(xe); > } > > -- > 2.46.0 >
On Wed, Sep 04, 2024 at 11:09:24AM +0530, Anshuman Gupta wrote: > On 2024-09-04 at 04:08:03 +0530, Rodrigo Vivi wrote: > > When going to d3cold we surely don't need to flush commits, > > nor handle fbdev. So, clean this up a bit. > > > > Also move the opregion call to make the function more symmetric. > > > > Cc: Imre Deak <imre.deak@intel.com> > > Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> > > --- > > drivers/gpu/drm/xe/display/xe_display.c | 12 +++--------- > > 1 file changed, 3 insertions(+), 9 deletions(-) > > > > diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c > > index daf9b9baa88a..3fd3e74f1310 100644 > > --- a/drivers/gpu/drm/xe/display/xe_display.c > > +++ b/drivers/gpu/drm/xe/display/xe_display.c > > @@ -310,9 +310,6 @@ static void xe_display_to_d3cold(struct xe_device *xe) > > * properly. > > */ > > intel_power_domains_disable(xe); > > - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_SUSPENDED, true); > How do we make sure fbconsole doesn't use vram to blink the cursor ? Good question, but that's not the point. The point is, if the fbconsole is there blinking the cursor, we are 100% sure that this runtime suspend call here won't ever happen. If we reach to this point. The fbconsole is not active and doesn't need to be suspended. > > Thanks, > Anshuman. > > - > > - xe_display_flush_cleanup_work(xe); > > > > intel_dp_mst_suspend(xe); > > > > @@ -329,18 +326,15 @@ static void xe_display_from_d3cold(struct xe_device *xe) > > > > intel_dmc_resume(xe); > > > > - drm_mode_config_reset(&xe->drm); > > - > > intel_display_driver_init_hw(xe); > > + > > + intel_opregion_resume(display); > > + > > intel_hpd_init(xe); > > > > /* MST sideband requires HPD interrupts enabled */ > > intel_dp_mst_resume(xe); > > > > - intel_opregion_resume(display); > > - > > - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_RUNNING, false); > > - > > intel_power_domains_enable(xe); > > } > > > > -- > > 2.46.0 > >
diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c index daf9b9baa88a..3fd3e74f1310 100644 --- a/drivers/gpu/drm/xe/display/xe_display.c +++ b/drivers/gpu/drm/xe/display/xe_display.c @@ -310,9 +310,6 @@ static void xe_display_to_d3cold(struct xe_device *xe) * properly. */ intel_power_domains_disable(xe); - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_SUSPENDED, true); - - xe_display_flush_cleanup_work(xe); intel_dp_mst_suspend(xe); @@ -329,18 +326,15 @@ static void xe_display_from_d3cold(struct xe_device *xe) intel_dmc_resume(xe); - drm_mode_config_reset(&xe->drm); - intel_display_driver_init_hw(xe); + + intel_opregion_resume(display); + intel_hpd_init(xe); /* MST sideband requires HPD interrupts enabled */ intel_dp_mst_resume(xe); - intel_opregion_resume(display); - - intel_fbdev_set_suspend(&xe->drm, FBINFO_STATE_RUNNING, false); - intel_power_domains_enable(xe); }
When going to d3cold we surely don't need to flush commits, nor handle fbdev. So, clean this up a bit. Also move the opregion call to make the function more symmetric. Cc: Imre Deak <imre.deak@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> --- drivers/gpu/drm/xe/display/xe_display.c | 12 +++--------- 1 file changed, 3 insertions(+), 9 deletions(-)