Able to reproduce the issue on Mac OS 10.11.4 using chrome latest Dev M52-52.0.2730.0. No overlay is observed on clicking attach file in Gmail.

Adding the Dev blocker as this is recent regression,Please feel free to remove if it's not.

CHANGELOG URL:
  https://chromium.googlesource.com/chromium/src/+log/f66d9059093dffbc634d997ff15afe9169916fdf..a75c88d82c63ab8109e5e0907e653bbb5ba0a6a6

Deleted: Actual_GmailAttachment.ogv 1.3 MB

Actual_GmailAttachment.ogv 1.3 MB Download

 Issue 610608  has been merged into this issue.

Suspecting: https://codereview.chromium.org/1956033002

mstensho@: Could you please take a look and confirm if the change is related.

Thank you!

I cannot reproduce this when making my own Chromium build (Linux). Built on this commit:

commit 814e1c24065cd6dd8e5d48a154ff3bf132c361d3
Author: chrome-cron <chrome-cron@google.com>
Date:   Mon May 9 03:11:57 2016 +0000

    Updating trunk VERSION from 2729.0 to 2730.0
    
    Cr-Commit-Position: refs/heads/master@{#392285}

I downloaded 52.0.2730.0 canary build for Windows, and I can see that it fails there, i.e. no file dialog is shown. Not sure if my patch has anything to do with it, though. Of course, when clicking a button, hit testing needs to work, but since there's visual feedback on the button when clicked, I don't know. Furthermore, if you tab-navigate to the button and hit enter, you also get visual feedback, but no file dialog. So hit-testing seems irrelevant.

So I doubt that this one is mine.

Could this be related to https://codereview.chromium.org/1945433002 then from the same regression range.

Assiging to rdevlin.cronin@ for further investigation.

I don't see any reason that https://codereview.chromium.org/1945433002 would have caused this, and it has also been reverted.  If the issue is still around, it was caused by something else.

Rebisect gives the same result as updated in C#3.

https://chromium.googlesource.com/chromium/src/+log/f66d9059093dffbc634d997ff15afe9169916fdf..a75c88d82c63ab8109e5e0907e653bbb5ba0a6a6

Below are the changes that remain with above 2 suspects already out of the window:
https://codereview.chromium.org/1940363002 (pasko@)
https://codereview.chromium.org/1959013002 (sigbjornf@)

This most likely looks to be pasko@'s change.

pasko@: Could you please take a look and confirm.

the change by pasko@ only modifies non-production code (tests) -> sigbjornf

Thanks, confirmed locally to be the source. I don't understand how that disposed timer unregistration would be observable, but reverting.

Reverted by https://chromium.googlesource.com/chromium/src/+/b0438b3ba5de4aa326db9197e86ed408edf710a2

Will re-land an adjusted version of that change, but wait until next week so as to let this revert settle in & verified as fixing the regression.

 Issue 611003  has been merged into this issue.

 Issue 611260  has been merged into this issue.