log and screenshot:
https://pantheon.corp.google.com/storage/browser/chromiumos-test-logs/bugfiles/cr/696048/?pli=1 

@helenzhang,

Do you also see the cursor streaking (trail is left visible?)

Yes. 

@helen has any bisecting been going on to figure out when this got introduced?

It's not reproducible in 9312.0.0/58.0.3015.0 on Wolf. 

reproduced on caroline R58-9334/ 58.0.3028

unable to reproduce on Caroline R58-9312/58.0.3015

reproduced on caroline with R58-9313/58.0.3021

+ marcheu@

Stephane, can you please have a look. This is currently blocking WM validation on Intel side.

I thought we had fixed this on the chrome side, can you retest with ToT chrome?

I don't see any changes in 3.18 kernel between those two version. There was a Chrome rev bump.

@marcheu reproduced on today's partner image (R59-9343/59.0.3030)

and today on R59-9344/59.0.3032 (still on caroline)

I tested R59-9344/59.0.3032 on a reef system (4.4 kernel) and could not reproduce the problem.

Stephane, since they are both using the same Chrome version doesn't sound like a Chrome issue. Caroline is on 3.18 and Reef on 4.4 kernel.

@marcheu, what fix did you think was in place? from Wayne's comment it looks like it might have fixed the world for 4.4, but not previous kernels?

What is the repro rate here? It sounds like this is a bit mixed.

This is currently marked as a beta blocker for 58 so we need to get this fixed soon if we are to promote next week. 

repro on 3.18 has been 100%, no idea on reef

We are going for beta promotion with the build Tuesday evening, any chance we can have a fix for this in that time?

Verified in 9334.40.0/58.0.3029.66 on Caroline.