[Intel-gfx] [PATCH] drm/i915: Only disable visible planes in intel_crtc_disable_planes

[Maarten Lankhorst]

This is fix for a regression introduced by 27321ae88c70104df
"drm/i915: Use the disable callback for disabling planes."

Disabling invisible planes may cause recalculation of
watermarks, which is a problem because the software state
is not yet in sync with the hardware state.
This may result in a black screen during kernel boot and
plymouth splash until any input action is performed in X.

Explicitly checking for plane visibility fixes the regression.
This is a patch for v4.2 only, v4.3 needs a different fix because
it was fixed by d032ffa04cf7c6f
"drm/i915: Handle disabling planes better, v2."

but later broken again in
4cf0ebbd4fafbdf "drm/i915: Rework plane readout."

This will be fixed in v4.3 by:
"drm/i915: Add .get_hw_state() method for planes"

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91952
Cc: João Paulo Rechi Vita <jprvita at endlessm.com>
Cc: stable at vger.kernel.org # v4.2 only
---
 drivers/gpu/drm/i915/intel_display.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 87476ff181dd..a5f97cfd86d8 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -4863,7 +4863,8 @@ static void intel_crtc_disable_planes(struct drm_crtc *crtc)
 
 	intel_crtc_dpms_overlay_disable(intel_crtc);
 	for_each_intel_plane(dev, intel_plane) {
-		if (intel_plane->pipe == pipe) {
+		if (intel_plane->pipe == pipe &&
+		    to_intel_plane_state(intel_plane->base.state)->visible) {
 			struct drm_crtc *from = intel_plane->base.crtc;
 
 			intel_plane->disable_plane(&intel_plane->base,
-- 
2.1.0