os: add fallback for undefined CPUs

[JungMinu]

Currently, for an unsupported OS, a call to os.cpus() throws an error within os.cpus() itself where it tries to get the length of undefined. This PR fixes the issue by adding fallback for undefined CPUs.

Fixes: #25483

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• commit message follows commit guidelines

[nodejs-github-bot]

@JungMinu sadly an error occured when I tried to trigger a build :(

[addaleax]

I’m wondering if it might make sense to add a “real” fallback, i.e. a dummy result that lists 1 CPU or so?

[devsnek]

I'd rather this threw. It's not accurate to return no cpus or one fake cpu. If this function can't find cpus it shouldn't make something up.

[addaleax]

@devsnek For context, this is a bugfix for a regression introduced in 51cea61. I think intentionally introducing a throw should be considered semver-major.

😢

[Trott]

@devsnek For context, this is a bugfix for a regression introduced in 51cea61. I think intentionally introducing a throw should be considered semver-major.

Does this affect any supported platforms?

Assuming this only affects unsupported platforms, I'd like to see us introduce the throw, but in a major release. (If it only affects unsupported platforms, I wouldn't consider it semver-major. But doing it in a major release anyway is the kind thing to do, so I'm 👍with that.) The behavior here will be difficult to test robustly unless it applies to a supported platform and configuration that we can test in CI. So removing the bits of code we can't test and that don't affect support platforms/configurations seems like the right thing to do.