Tested on Windows 10:

Chrome Stable 62.0.3202.89 (Official Build) (64-bit) (cohort: 62_89_win): BAD
Chrome Canary 64.0.3266.0 (Official Build) canary (32-bit) (cohort: Clang-32): GOOD

Requesting reverse bisect.

Able to reproduce this issue on Mac OS 10.12.6, Windows 7 and Ubuntu 14.04 on latest stable 62.0.3202.89 with the steps mentioned in the original comment.
Issue is not observed on latest Canary 64.0.3266.0, Dev 64.0.3260.2 and Beta 63.0.3239.40.

Performed Reverse Bisect and below is the bisect information.

Reverse Bisect Information:
=====================
Good build: 63.0.3239.27 
Bad Build : 63.0.3239.26 

As both good and bad are branch builds, unable to run the per-revision bisect script. Hence providing the Change log URL through Manual Bisect from Omahaproxy .

Change Log URL: 
===============
https://chromium.googlesource.com/chromium/src/+log/63.0.3239.26..63.0.3239.27?pretty=fuller&n=10000

From the above change log the possible CL suspect is:
Reviewed-on: https://chromium-review.googlesource.com/740061

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

Adding ReleaseBlock-Stable label as this issue is seen on Stable build. Please feel free to remove the same if it is not applicable.


Thanks...

I don't believe this feature was introduced until 63:

https://www.chromestatus.com/features/5734614437986304

Overscroll-behavior does not exist in M62. So this only affects M63.
Also I don't think this is a major issue to be a release blocker. So removing
that label and making this s P2.


Assigning to sunyunjia@ to investigate.


Can this be related to how we 

BTW, I removed the google restricted view since it is not really necessary. I think it was added since the bug was filed with @google.com email address.

Looking at this change log: https://chromium.googlesource.com/chromium/src/+log/63.0.3239.26..63.0.3239.27?pretty=fuller&n=10000

A renaming from scroll-boundary-behavior to overscroll-behavior happens during this period. So this should just a issue due to renaming. The renaming has been merged to M63, so the problem won't really affect its being shipped.

That video clip shows the demo page from ebidelman@. I am pretty sure that uses
the correct property name. Also it shows the property working in some situation
but not the other. If this was a naming issue then it wouldn't work at all.

I think the bisect in #2 is probably not accurate because you can recreate the
problem even when the feature is not available (pre M62 or early M63 before rename). 
So it can lead to incorrect bad revision.

sunyunjia@: have you tried recreating the issue yourself on Canary?

 

Sorry my bad. The background won't bounce if it is a touch-move scroll, but will bounce if it is a fling scroll. Let me take a look at it.