Merge tag 'platform-drivers-x86-v4.12-2' of git://git.infradead.org/linux-platform...

[karo-tx-linux.git] / drivers / gpu / drm / i915 / intel_display.c

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 3cabe52a4e3b168e176d1f55abdae65f67219ef7..569717a1272367a91cf682a9cae7640f9ae32777 100644 (file)
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -12203,6 +12203,15 @@ static void update_scanline_offset(struct intel_crtc *crtc)
         * type. For DP ports it behaves like most other platforms, but on HDMI
         * there's an extra 1 line difference. So we need to add two instead of
         * one to the value.
+        *
+        * On VLV/CHV DSI the scanline counter would appear to increment
+        * approx. 1/3 of a scanline before start of vblank. Unfortunately
+        * that means we can't tell whether we're in vblank or not while
+        * we're on that particular line. We must still set scanline_offset
+        * to 1 so that the vblank timestamps come out correct when we query
+        * the scanline counter from within the vblank interrupt handler.
+        * However if queried just before the start of vblank we'll get an
+        * answer that's slightly in the future.
         */
        if (IS_GEN2(dev_priv)) {
                const struct drm_display_mode *adjusted_mode = &crtc->config->base.adjusted_mode;