99th_percentile_frame_build_time_millis regression...

[kf6gpe]

Graph: https://flutter-flutter-perf.skia.org/e/?begin=1620771842&end=1620856872&keys=Xfa7e0369df3d4d1a6eebae659dd05ade&num_commits=50&request_type=1&xbaroffset=24022.

Clear signal at or around the roll at #82322.

/cc @jason-simmons, this week's engine sheriff.

[jason-simmons]

This started with flutter/engine@8e72d7b

I noticed that the call to EngineLayer::AssociateWithDartWrapper that now happens within SceneBuilder::addPicture is taking tens of milliseconds on a few outlier frames in this benchmark
(flutter/engine@8e72d7b#diff-652a2a4e7066572a99d399deb9c7b8168e42f647481c028d106e6dc84e04f2b1R31)

@dnfield

[dnfield]

That seems really strange, but we should just revert the change for now in that case. I may have an alternative solution that doesn't require it.

[jason-simmons]

Specifically the time seems to be attributed to a call to Dart_NewWeakPersistentHandle. Perhaps the change to addPicture can result in some additional GCs happening during frame builds?

AFAIK there was no significant change to other benchmarks, so it's possible that this is an atypical workload.

[dnfield]

https://flutter-flutter-perf.skia.org/e/?begin=1619599741&end=1620777543&queries=sub_result%3Dnew_gen_gc_count%26sub_result%3Dold_gen_gc_count%26test%3Dlinux_cull_opacity_perf__e2e_summary%26test%3Dmac_android_cull_opacity_perf__timeline_summary&requestType=0&showZero=false&xbaroffset=24022 shows some more GCs happening but not many more than happened ont he previous run.

[github-actions]

This thread has been automatically locked since there has not been any recent activity after it was closed. If you are still experiencing a similar issue, please open a new bug, including the output of flutter doctor -v and a minimal reproduction of the issue.