All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=733994

Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?keys=agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDgtoKy9AoM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg9viG9woM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg1ueE9woM,agxzfmNocm9tZXBlcmZyFAsSB0Fub21hbHkYgIDg1uP7swsM


Bot(s) for this bug's original alert(s):

webrtc-android-tests-nexus4-lollipop
webrtc-android-tests-nexus72

I don't follow what Chromium roll we're talking about here. I can only see 3 regular WebRTC commits in the blame list:
https://chromium.googlesource.com/external/webrtc/+log/0ef8fb91b4930783f0f32cbd3a68e5f98b6ebdd4..8fa21c49eff52b1848ff16483cc3a07af6acc7ca

I know that Android switched from GCC -> Clang in #18537 (https://codereview.webrtc.org/2930243004/) which caused performance changes, but that's far earlier than this.

If we're sure a Clang roll is a perf regression we should probably follow up, yes. First thing would be to look for similar bugs in Chromium though.

This is the blame list I get. Did I misunderstand something?

https://chromium.googlesource.com/external/webrtc/+log/7dce727c88e508f62b88aa68715fd5d6967bfbf4..b7d601552971992d40e84a70f936ea4e38044d91

Sorry, my bad (clicked the wrong alert in the graph).
You're right, it's a regular Clang roll, and it can affect Android since it's after they switched to use Clang... so if you think the regression is enough to follow up on - file a clang bug or dig into similar Chromium bugs in that range.