SIGSEGV on PhaseIdealLoop::split_up?

[Dawid Weiss]

> [...] But as the issues we see don't happen on the inner virtual machines that remove some advanced CPU features, maybe that's special here.

Yes - I meant a particular hardware configuration your setup runs on.
I think we're safe to rule out hardware failure as all stack traces
are (nearly) the same.

> The recent failures don't happen with Java 8 and Java 9, but started with Java 10 or later! (because we only see the bug on runs using those versions).

They do occur with Java 9 (see mailing list archives) and stretch as
far back as July 10th, 2018. They are rare though, given how many test
runs we've had since then, so it'll be a tricky one. ;) My point about
not being able to reproduce the issue has more to do with the fact
that the problem may not be repeatable in isolation from other,
previously executed, tests that prime execution statistics for the
optimizer. But I haven't tried, so it's just a gut feeling.


D.