Able to reproduce this issue on Latest Canary#61.0.3160.0 for 'Win7'.

Able to reproduce the issue and is broken in M51 on Windows, Mac & Ubuntu 14.04. Issue is observed on Stable 59.0.3071.115 and as per above comment on canary as well. Below are the bisect details for the same:

Bisect info:
============
51.0.2686.0 - Good Build
51.0.2687.0 - Bad Build

Bisect URL:
===========
https://chromium.googlesource.com/chromium/src/+log/6bebadddb399ad1c1f91592efed410040617e708..5840ac85a7c912d9ddb764007b6a9ec65a26ad22

Suspecting below change could be a possible culprit:

Change-log:
==========
https://chromium.googlesource.com/chromium/src/+/8796ca1186b3ff3e3da105b3ee41f1698be01821

@ mstensho: Assigning to you, kindly take a look into it. Please help us to find an owner if not with respect to your change.

Thanks.!

I can scroll it just fine with the arrow keys on the keyboard, but not with the scroll wheel on the mouse. I'm assuming this is the issue? Otherwise, I misunderstood.

Attaching a demo. Please verify that this is the issue you're talking about. Removing visibility:hidden on the scrollable container makes the problem go away, FWIW.

Deleted: demo.html 348 bytes

demo.html 348 bytes View Download

Sometimes it is possible to scroll this with the arrow keys and sometimes its not, but I don't know why.

I've done a little bit of investigation and figured out it is sometimes possible to scroll even with the mouse wheel. But if it is possible, its very limited (e.g. it works for about 2 seconds and then the functinality is just gone again). The same counts for the arrow keys.
It worked only with the developer tools open so far. It worked more often after I've resized the window and scrolled immediately after. 

I really don't think the blamed CL is the one, though. It's purely multicol specific, and there's no multicol in the tests here.

https://chromium.googlesource.com/chromium/src/+/ca4feba9a31e10d1baab5813650c192e06cfd643 is probably a better candidate. CCing author and a couple of scrolling experts in Blink (since I suspect the root cause is in Blink). Lowering priority, since it's a very old regression. Unassigning myself.

flackr@, smcgruer@ please triage. This is another issue related to threaded scrolling that occurs but really shouldn't.

This seems to belong on the list of effects of hit testing/ event routing failures.

flackr to group and redirect...

It is probably the case that the hittest is not hitting because the element is hidden. The other browsers must be special casing this in some way.

flackr, any update on this issue?

flackr@ is still OOO for another week.

However I cannot reproduce this using the original fiddle on Ubuntu 71.0.3578.98. nzolghadr@, can you confirm if you can reproduce (I may just be doing it wrong).

I will run a when-was-it-fixed bisect later today but would appreciate confirmation of whether others can reproduce or not.

I tested on Mac and Windows as well and I am able to scroll with mouse wheel.

Hrm, when bisecting I saw that it seems improved but not perfect. There is a definite jump from when it doesn't move at all to when it moves almost all the time, but I was able to reproduce a few cases where it seemed to get 'stuck' at the top of the scroll for multiple seconds before I was suddenly able to wheel-scroll again.

The bisect also pulled out a very strange CL:

You are probably looking for a change made after 552588 (known bad), but no later than 552589 (first known good).
CHANGELOG URL:
  https://chromium.googlesource.com/chromium/src/+log/7bbdb70b2643201e9ec6236443b7b2003deaf03d..fb1ccf02ee8ca79e1404abfd3a3a7d540b7d2dbd

https://chromium.googlesource.com/chromium/src/+/fb1ccf02ee8ca79e1404abfd3a3a7d540b7d2dbd

'Make --site-per-process the default on ToT via fieldtrial_testing_config'

Confirmed that using --disable-site-isolation-trials reproduces the original broken behavior. 

So this can still be reproduced when the scroll target *isn't* out of process (interesting!) e.g. at https://output.jsbin.com/jataxag

It seems the scroller is not composited, looks like blink side problem.