Talked with Ken and he will handle the revert.

Revert occurring in https://codereview.chromium.org/2138263002/ .

Verified that https://build.chromium.org/p/chromium.perf/builders/Win%207%20Perf%20%284%29/builds/6755 , which contains the revert above, and subsequent builds on that machine are green.

We should figure out how to write a more targeted test when attempting to re-land this change.

I'm trying to figure out how to repro now. Seems like this will be quite
difficult.

I've managed to repro hangs locally under heavy CPU load. They aren't
actually caused by the CL, but are significantly exacerbated by it due to
changes in timing. I'll have a fix for the real bug out soon, and then I'd
like to try relanding in off-peak hours assuming the local repro disappears.

That's great. Thanks Ken for persevering with this. See also  Issue 628765 , where random browser hangs upon startup are seen on multiple platforms. I wonder whether these issues might have the same root cause. Please go ahead with your re-land when you're ready, and please watch the Perf bots afterward, since they were the ones most reliably affected. Thanks.

Yeah, I checked 628765 earlier hoping it might have been my fault too. Unfortunately it looks like the those hangs are still occurring after my fix landed.

Fix is in r405776.

This failure https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Debug%20%28New%20Intel%29/builds/2847 was built from r405822, and this one https://build.chromium.org/p/chromium.gpu.fyi/builders/Mac%20Retina%20Release/builds/5618 was built from r405805.

I do however have a Windows machine crunching non-stop on starting new browser processes under very heavy CPU load, in the hopes that I might repro the hangs anyway. Maybe I should spin up a Mac too.