New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 855217 link

Starred by 2 users

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

Figure out why Chrome still uses two surfaces in fullscreen overlay mode

Project Member Reported by zmo@chromium.org, Jun 21 2018

Issue description

Feedback from Intel colleagues, even when Chrome is playing a video with nothing on top (overlay mode), they still see two surfaces (one BGRA, one YUY2) are getting rendered. As comparison, in Edge in the same mode, they only see one surface.

We need to figure out why the BGRA surface is involved and get rid of that.
 
I double checked this, and it seems when damage is zero we skip the root render pass completely. This means we don't call SetDrawRectangle, BeginDraw, EndDraw, or Commit. I've verified this behavior by logging locally. I've also inspected DirectRenderer:

https://cs.chromium.org/chromium/src/components/viz/service/display/direct_renderer.cc?rcl=001e0436c511034494b32f410e01bbd9fd8f6dfa&l=321

  // We can skip all drawing if the damage rect is now empty.
  bool skip_drawing_root_render_pass =
      current_frame()->root_damage_rect.IsEmpty() && allow_empty_swap_;

allow_empty_swap_ is true when using DirectComposition.

All of the above assumes that we hit the overlay path where nothing is on top of the video. If there are any controls or UI on top of the video, then we fall back to compositing because we haven't enabled underlays yet.

Can we get clarification from Intel about the exact testing scenario, and specifically if they had controls or UI like "press Esc to exit fullscreen" on top of the video when playing?
I just emailed Intel colleagues and hopefully we can clarify this and move forward. 
I replied back with the appropriate people for testing.  I will retest this week and provide more details.
Cc: akepati....@intel.com richard....@intel.com
Labels: Needs-Feedback
Richard, did you get a chance to take a look at this?
Labels: -Pri-1 Pri-2
[GPU Triage Council]

Sign in to add a comment