This test is flaking on a variety of bots:

https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=content_browsertests&tests=SitePerProcessBrowserTouchActionTest

Here are results from a particular build:

https://chromium-swarm.appspot.com/task?id=400f935166248b10&refresh=10&show_raw=1

where the log is:

[1839/2933] SitePerProcessBrowserTouchActionTest.EffectiveTouchActionPropagatesAcrossFrames (3147 ms)
[0920/045635.524555:ERROR:kill_posix.cc(84)] Unable to terminate process group 12969: No such process (3)
[ RUN      ] SitePerProcessBrowserTouchActionTest.EffectiveTouchActionPropagatesAcrossNestedFrames
Xlib:  extension "RANDR" missing on display ":99".

DevTools listening on ws://127.0.0.1:33981/devtools/browser/2d15adbc-fb64-479a-9978-b83f1c6c161c
../../content/browser/site_per_process_browsertest.cc:12470: Failure
Expected equality of these values:
  expected_touch_action
    Which is: 63
  whitelisted_touch_action.value()
    Which is: 15

Looking at 

https://cs.chromium.org/chromium/src/content/browser/site_per_process_browsertest.cc?rcl=68954ea7cf6d9a7a56e0bfa6e8417bcca3466fa5&l=12292

the returned value is probably just not available, i.e. the optional value has no value.

It seems like a bad design to use the return value from

GetEffectiveTouchActionForChild()

in an EXPECT_EQ() comparison, given that the value it returns may or may not have a meaningful value.