drm/amd/display: Handle commit plane with no FB.

1) Always call flip ISR, for any surface comit, not only for page flip.

2) For events not signaled in flip ISR, handle them in end of commit_tail.

3)Fix race condition between subsequent atomic calls by removing
  current state access AFTER drm_atomic_helper_commit_hw_done
  was called. After this call a pending next commit will
  swap in his own state into objects and proceed.
Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -222,15 +222,15 @@ static void dm_pflip_high_irq(void *interrupt_params)
 
 
 	/* wakeup usersapce */
-	if (amdgpu_crtc->event
-			&& amdgpu_crtc->event->event.base.type
-			== DRM_EVENT_FLIP_COMPLETE) {
+	if (amdgpu_crtc->event) {
 		/* Update to correct count/ts if racing with vblank irq */
 		drm_crtc_accurate_vblank_count(&amdgpu_crtc->base);
 
 		drm_crtc_send_vblank_event(&amdgpu_crtc->base, amdgpu_crtc->event);
+
 		/* page flip completed. clean up */
 		amdgpu_crtc->event = NULL;
+
 	} else
 		WARN_ON(1);
 

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_types.c

@@ -2180,6 +2180,24 @@ static void handle_cursor_update(
 }
 
 
+static void prepare_flip_isr(struct amdgpu_crtc *acrtc)
+{
+
+	assert_spin_locked(&acrtc->base.dev->event_lock);
+	WARN_ON(acrtc->event);
+
+	acrtc->event = acrtc->base.state->event;
+
+	/* Set the flip status */
+	acrtc->pflip_status = AMDGPU_FLIP_SUBMITTED;
+
+	/* Mark this event as consumed */
+	acrtc->base.state->event = NULL;
+
+	DRM_DEBUG_DRIVER("crtc:%d, pflip_stat:AMDGPU_FLIP_SUBMITTED\n",
+						 acrtc->crtc_id);
+}
+
 /*
  * Executes flip
  *
@@ -2249,17 +2267,8 @@ static void amdgpu_dm_do_flip(
 	addr.flip_immediate = async_flip;
 
 
-	if (acrtc->base.state->event &&
-	    acrtc->base.state->event->event.base.type ==
-			    DRM_EVENT_FLIP_COMPLETE) {
-		acrtc->event = acrtc->base.state->event;
-
-		/* Set the flip status */
-		acrtc->pflip_status = AMDGPU_FLIP_SUBMITTED;
-
-		/* Mark this event as consumed */
-		acrtc->base.state->event = NULL;
-	}
+	if (acrtc->base.state->event)
+		prepare_flip_isr(acrtc);
 
 	surface_updates->surface = dc_stream_get_status(acrtc->stream)->surfaces[0];
 	surface_updates->flip_addr = &addr;
@@ -2274,8 +2283,6 @@ static void amdgpu_dm_do_flip(
 
 
 	spin_unlock_irqrestore(&crtc->dev->event_lock, flags);
-	DRM_DEBUG_DRIVER("crtc:%d, pflip_stat:AMDGPU_FLIP_SUBMITTED\n",
-						 acrtc->crtc_id);
 }
 
 static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
@@ -2289,13 +2296,13 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 	struct drm_plane_state *old_plane_state;
 	const struct dc_stream *dc_stream_attach;
 	const struct dc_surface *dc_surfaces_constructed[MAX_SURFACES];
+	struct amdgpu_crtc *acrtc_attach = to_amdgpu_crtc(pcrtc);
 	int planes_count = 0;
 
 	/* update planes when needed */
 	for_each_plane_in_state(state, plane, old_plane_state, i) {
 		struct drm_plane_state *plane_state = plane->state;
 		struct drm_crtc *crtc = plane_state->crtc;
-		struct amdgpu_crtc *acrtc_attach = to_amdgpu_crtc(crtc);
 		struct drm_framebuffer *fb = plane_state->fb;
 		struct drm_connector *connector;
 		struct dm_connector_state *con_state = NULL;
@@ -2306,7 +2313,7 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 			continue;
 		}
 
-		if (!fb || !crtc || !crtc->state->active)
+		if (!fb || !crtc || pcrtc != crtc || !crtc->state->active)
 			continue;
 
 		pflip_needed = !state->allow_modeset;
@@ -2339,7 +2346,7 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 				continue;
 
 
-			if (crtc == pcrtc) {
+
 			add_surface(dm->dc, crtc, plane,
 				    &dc_surfaces_constructed[planes_count]);
 			if (dc_surfaces_constructed[planes_count] == NULL) {
@@ -2348,8 +2355,12 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 			}
 			dc_stream_attach = acrtc_attach->stream;
 			planes_count++;
-			}
+
 		} else if (crtc->state->planes_changed) {
+			/* Assume even ONE crtc with immediate flip means
+			 * entire can't wait for VBLANK
+			 * TODO Check if it's correct
+			 */
 			*wait_for_vblank =
 				acrtc_attach->flip_flags & DRM_MODE_PAGE_FLIP_ASYNC ?
 				false : true;
@@ -2359,6 +2370,8 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 				fb,
 				drm_crtc_vblank_count(crtc) + *wait_for_vblank);
 
+			/*TODO BUG remove ASAP in 4.12 to avoid race between worker and flip IOCTL */
+
 			/*clean up the flags for next usage*/
 			acrtc_attach->flip_flags = 0;
 		}
@@ -2366,15 +2379,27 @@ static void amdgpu_dm_commit_surfaces(struct drm_atomic_state *state,
 	}
 
 	if (planes_count) {
+		unsigned long flags;
+
+		if (pcrtc->state->event) {
+
+			drm_crtc_vblank_get(pcrtc);
+
+			spin_lock_irqsave(&pcrtc->dev->event_lock, flags);
+			prepare_flip_isr(acrtc_attach);
+			spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
+		}
+
 		if (false == dc_commit_surfaces_to_stream(dm->dc,
 							  dc_surfaces_constructed,
 							  planes_count,
-							  dc_stream_attach)) {
+							  dc_stream_attach))
 			dm_error("%s: Failed to attach surface!\n", __func__);
-			return;
-		}
+
 		for (i = 0; i < planes_count; i++)
 			dc_surface_release(dc_surfaces_constructed[i]);
+	} else {
+		/*TODO BUG Here should go disable planes on CRTC. */
 	}
 }
 
@@ -2405,7 +2430,6 @@ void amdgpu_dm_atomic_commit_tail(
 		struct drm_crtc_state *new_state = crtc->state;
 
 		acrtc = to_amdgpu_crtc(crtc);
-
 		aconnector =
 			amdgpu_dm_find_first_crct_matching_connector(
 				state,
@@ -2571,14 +2595,6 @@ void amdgpu_dm_atomic_commit_tail(
 				dc_stream_get_status(acrtc->stream)->primary_otg_inst;
 	}
 
-	/* update planes when needed per crtc*/
-	for_each_crtc_in_state(state, pcrtc, old_crtc_state, j) {
-		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(pcrtc);
-
-		if (acrtc->stream)
-			amdgpu_dm_commit_surfaces(state, dev, dm, pcrtc, &wait_for_vblank);
-	}
-
 	for (i = 0; i < new_crtcs_count; i++) {
 		/*
 		 * loop to enable interrupts on newly arrived crtc
@@ -2592,20 +2608,28 @@ void amdgpu_dm_atomic_commit_tail(
 		manage_dm_interrupts(adev, acrtc, true);
 	}
 
+	/* update planes when needed per crtc*/
+	for_each_crtc_in_state(state, pcrtc, old_crtc_state, j) {
+		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(pcrtc);
 
-	/*TODO mark consumed event on all crtc assigned event
-	 * in drm_atomic_helper_setup_commit just to signal completion
+		if (acrtc->stream)
+			amdgpu_dm_commit_surfaces(state, dev, dm, pcrtc, &wait_for_vblank);
+	}
+
+
+	/*
+	 * send vblank event on all events not handled in flip and
+	 * mark consumed event for drm_atomic_helper_commit_hw_done
 	 */
 	spin_lock_irqsave(&adev->ddev->event_lock, flags);
 	for_each_crtc_in_state(state, crtc, old_crtc_state, i) {
 		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
 
-		if (acrtc->base.state->event &&
-				acrtc->base.state->event->event.base.type != DRM_EVENT_FLIP_COMPLETE) {
-			acrtc->event = acrtc->base.state->event;
+		if (acrtc->base.state->event)
+			drm_send_event_locked(dev, &crtc->state->event->base);
+
 		acrtc->base.state->event = NULL;
 	}
-	}
 	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
 
 	/* Signal HW programming completion */
@@ -2614,23 +2638,6 @@ void amdgpu_dm_atomic_commit_tail(
 	if (wait_for_vblank)
 		drm_atomic_helper_wait_for_vblanks(dev, state);
 
-	/*TODO send vblank event on all crtc assigned event
-	 * in drm_atomic_helper_setup_commit just to signal completion
-	 */
-	spin_lock_irqsave(&adev->ddev->event_lock, flags);
-	for_each_crtc_in_state(state, crtc, old_crtc_state, i) {
-		struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
-
-		if (acrtc->event &&
-			acrtc->event->event.base.type != DRM_EVENT_FLIP_COMPLETE) {
-			drm_send_event_locked(dev, &acrtc->event->base);
-			acrtc->event = NULL;
-		}
-	}
-	spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
-
-	/*TODO Is it to early if actual flip haven't happened yet ?*/
-	/* Release old FB */
 	drm_atomic_helper_cleanup_planes(dev, state);
 }