Cirrus CI: nightly freebsd-13 x86_64 is quite flaky

[JohnTitor]

That CI env is currently quite fragile and fails at almost time.

The failure looks like (link):

failures:

---- test_entrypoint_macro stdout ----

thread 'test_entrypoint_macro' (103040) panicked at ctest/tests/basic.rs:60:71:
called `Result::unwrap()` on an `Err` value: MacroExpansion("tests/input/macro.rs", "macro expansion failed with signal: 11 (SIGSEGV) (core dumped): ")
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

It happens on random tests.
It'd be great if we could avoid it by tweaking CI config or something else.

Related: #4641

[brad0]

From watching a bunch of CI builds it was i686 not x86_64.

[JohnTitor]

It means the issue wasn't specific to x86_64, thank you.
My current guess is the resource limitation is involved somehow.

[brad0]

It means the issue wasn't specific to x86_64, thank you.

That's a possibility.

I remember seeing this PR..

#4641

[JohnTitor]

Ah interesting, didn't catch up that PR.

[brad0]

Ya, clicking on commits I see some i686, but definitely more x86_64.

[tgross35]

I don't think there is anything actionable on our end here since the intermediate segfaults seem to be a kernel bug https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=285873. Since FreeBSD 13 is (aiui) approaching EOL, it's probably not worth too much effort trying to work around it - I've just been rerunning that job when it fails on FreeBSD-related PRs.