Ah yes, I know what's going on.  It's definitely my merge (https://chromium.googlesource.com/chromium/src/+/278eae56232e3c2c8d36326f7969370c7d43ae91) -- I didn't realize that the test I merged depended on a behavior change in a test helper function NavigateFrameToURL (wait for full load, rather than commit), which was introduced in https://codereview.chromium.org/1816733002/ and isn't in M50.  I can fix it by adding a WaitForLoadStop() after that call.  It's also fine to disable to test in the meantime -- this doesn't indicate a problem with the fix itself.

Proposed fix: https://codereview.chromium.org/1900923003/

Just to make sure - is it ok to merge the fix from #2 to M50?  The test I merged in  issue 576204  will be flaky without it.

Has this CL made it to ToT and validated there?

#5: to clarify, this test fix is only needed in M50. It is not checked in to ToT, as it is not needed there.

Here's what has happened so far:
- there was a sandbox flags bug ( issue 576204 ) which I fixed in https://codereview.chromium.org/1868823002.  That CL also added a test for the bug.
- I verified that fix in Canary and got permission to merge to M50.
- I merged the fix and the new test to M50 in https://codereview.chromium.org/1899843002.
- This bug was filed, as the new test turned out to be flaky on M50 (but fine on ToT) -- see explanation in #1.
- The fix in #2 would fix that flakiness on M50.

Seeing same failure on Mac stable(50.0.2661.86) on Official Desktop Continuous builder as well:

Link to the builder:
====================
https://uberchromegw.corp.google.com/i/official.desktop.continuous/builders/mac%20stable/builds/586

Link to the log:
=================
https://uberchromegw.corp.google.com/i/official.desktop.continuous/builders/mac%20stable/builds/586/steps/content_browsertests/logs/stdio

alexmos@: Could you please take a look at this and confirm if these failures are same as seen on Linux.

Thank you!

[Automated comment] Request affecting a post-stable build (M50), manual review required.

#7: yes, this is the same failure as the one on Linux.  It should also be fixed by the CL in #2 (https://codereview.chromium.org/1900923003/) once I get permission to land it to 2661.

Issue 605078 has been merged into this issue.

Merge approved for M50 branch 2661, assuming we're only tweaking test code with no functional changes.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/ff228a374881cbd34de7a4c15c12681679de2fe0

commit ff228a374881cbd34de7a4c15c12681679de2fe0
Author: Alex Moshchuk <alexmos@chromium.org>
Date: Thu Apr 21 23:02:22 2016

Fix CrossProcessPopupInheritsSandboxFlagsWithNoOpener on M50 official builds.

The test depended on a behavior change in NavigateFrameToURL (wait for
full load, rather than commit), which was introduced in
https://codereview.chromium.org/1816733002/ and isn't in M50.  Use
WaitForLoadStop to wait for the full page to load instead.

BUG= 604559 
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation
R=creis@chromium.org

Review URL: https://codereview.chromium.org/1900923003 .

Cr-Commit-Position: refs/branch-heads/2661@{#620}
Cr-Branched-From: ef6f6ae5e4c96622286b563658d5cd62a6cf1197-refs/heads/master@{#378081}

[modify] https://crrev.com/ff228a374881cbd34de7a4c15c12681679de2fe0/content/browser/frame_host/render_frame_host_manager_browsertest.cc

[Android]
Same failure observed on latest M50 build : 

Build bot link:
https://chromegw.corp.google.com/i/official.android/builders/test-official-arm/builds/508

Content_browsertests failure log:
https://chromegw.corp.google.com/i/official.android/builders/test-official-arm/builds/508/steps/Unit%20Test%3A%20content_browsertests/logs/stdio


[==========] 1020 tests ran.
C  872.127s Main  [  PASSED  ] 1019 tests.
C  872.127s Main  [  FAILED  ] 1 test, listed below:
C  872.127s Main  [  FAILED  ] RenderFrameHostManagerTest.CrossProcessPopupInheritsSandboxFlagsWithNoOpener (UNKNOWN)
C  872.127s Main  
C  872.127s Main  1 FAILED TEST


adding current sheriffs 

Please don't cc sheriffs if you just need fixes to be merged on branches, esp. if the appropriate people are already in the loop.

This should be fixed.  I don't see the failure anymore on the linked builders.