[net_unittests] Add 1µs delay for cookie timestamps - #5216

@kaidokert where do we stand on filtering tests vs making changes to upstream files? This change doesn't seem like a good idea IMO

@kaidokert where do we stand on filtering tests vs making changes to upstream files? This change doesn't seem like a good idea IMO

Thanks for the feedback. I'd like to add that this test can potentially fail on other high-performance platforms as well, due to rapid execution, not just on Xbox.

@kaidokert where do we stand on filtering tests vs making changes to upstream files? This change doesn't seem like a good idea IMO

Couple of comments on this

• This is on a maintenance branch, i'm less concerned about upstream changes here, as it's not actively increasing the diff in our mainline code

• The Cookiemonster timing tests have had a lot of timing related flakiness issues for both us and Chromium, it looks like the test code is actually broken/unreliable. Chromium bug tracker has a lot of similar flakiness issues

• The workaround looks like a sensible way to get around it

• Ideally we should send a patch to Chromium upstream as well, the code is unchanged in Chromium trunk

• I think an even cleaner fix would be to update the whole test suite to use simulated timer in a task environment, e.g. base::test::TaskEnvironment(base::test::TaskEnvironment::TimeSource::MOCK_TIME) and FastForwardBy etc. or base::SimpleTestClock. That's a larger rewrite though

As is, i would land this to remove the flakiness source.