drm/i915: Always recompute watermarks when distrust_bios_wm is set, v2.

On some systems there can be a race condition in which no crtc state is
added to the first atomic commit. This results in all crtc's having a
null DDB allocation, causing a FIFO underrun on any update until the
first modeset.

Changes since v1:
- Do not take the connection_mutex, this is already done below.

Reported-by: Maarten Lankhorst <[email protected]>
Inspired-by: Mahesh Kumar <[email protected]>
Signed-off-by: Maarten Lankhorst <[email protected]>
Fixes: 98d3949 ("drm/i915/gen9: Compute DDB allocation at atomic
check time (v4)")
Cc: <[email protected]> # v4.8+
Cc: Mahesh Kumar <[email protected]>
Cc: Matt Roper <[email protected]>
Link: http://patchwork.freedesktop.org/patch/msgid/[email protected]
Reviewed-by: Mahesh Kumar <[email protected]>
Reviewed-by: Matt Roper <[email protected]>

drivers/gpu/drm/i915/intel_pm.c

@@ -4819,10 +4819,18 @@ skl_compute_wm(struct drm_atomic_state *state)
 	struct drm_crtc_state *cstate;
 	struct intel_atomic_state *intel_state = to_intel_atomic_state(state);
 	struct skl_wm_values *results = &intel_state->wm_results;
+	struct drm_device *dev = state->dev;
 	struct skl_pipe_wm *pipe_wm;
 	bool changed = false;
 	int ret, i;
 
+	/*
+	 * When we distrust bios wm we always need to recompute to set the
+	 * expected DDB allocations for each CRTC.
+	 */
+	if (to_i915(dev)->wm.distrust_bios_wm)
+		changed = true;
+
 	/*
 	 * If this transaction isn't actually touching any CRTC's, don't
 	 * bother with watermark calculation.  Note that if we pass this
@@ -4833,6 +4841,7 @@ skl_compute_wm(struct drm_atomic_state *state)
 	 */
 	for_each_new_crtc_in_state(state, crtc, cstate, i)
 		changed = true;
+
 	if (!changed)
 		return 0;