Able to reproduce the issue on 63.0.3218.0, latest canary 63.0.3222.0 and on latest stable 61.0.3163.100 using windows 10, Ubuntu 14.04 and Mac 10.12.6 with the steps mentioned in original comment.

Below is the Bisect Information:

Good build: 53.0.2762.0
Bad Build: 53.0.2763.0

You are probably looking for a change made after 398545 (known good), but no later than 398595 (first known bad).
CHANGELOG URL:

https://chromium.googlesource.com/chromium/src/+log/67a46a72b12d0a556cf0ccec546b08be058771ea..dd89a1e1f3f9e283d157f4edb9d419c899f36bf4

From the above change log Suspecting: https://codereview.chromium.org/2046883003

@jbroman: Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Note: Getting long list of CLs with per-revision . bisect script as well.

Thanks!!

There's unlikely to be a straightforward fix to this; bouncing to paint team for triage (as I'm no longer working on paint).

Regressions from M-53 are not regressions for bug tracking purposes. jbroman@ is right; this is not likely to be an easy fix.

As there is easy workaround, I would like to leave this to SPv2 to fix.

What's the exact order of -webkit-box-reflect? After mask/clip-path, before opacity/mix-blend-mode?

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot