Makes render time > decode time in VideoFrameMatcher.
Without this, we can end up with negative capture-to-render delays
if the jitter buffer sets the render time to an earlier time.
Bug: webrtc:11017
Change-Id: I590509136f630d025cde6e5e13d4a3ee620267ae
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/156081
Commit-Queue: Sebastian Jansson <srte@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29409}
diff --git a/test/scenario/video_frame_matcher.cc b/test/scenario/video_frame_matcher.cc
index d2b0ee2..2f95b29 100644
--- a/test/scenario/video_frame_matcher.cc
+++ b/test/scenario/video_frame_matcher.cc
@@ -112,7 +112,9 @@
frame_pair.decode_id = captured.best_decode->id;
frame_pair.decoded = captured.best_decode->frame;
frame_pair.decoded_time = captured.best_decode->decoded_time;
- frame_pair.render_time = captured.best_decode->render_time;
+ // We can't render frames before they have been decoded.
+ frame_pair.render_time = std::max(captured.best_decode->render_time,
+ captured.best_decode->decoded_time);
frame_pair.repeated = captured.best_decode->repeat_count++;
}
for (auto& handler : frame_pair_handlers_)
