src: handle empty Maybe in uv binding initialize

[addaleax]

This can fail when terminating a Worker that loads
the uv binding at the same time.

Refs: #25061 (comment)

Checklist

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

[nodejs-github-bot]

@addaleax build started: https://ci.nodejs.org/blue/organizations/jenkins/node-test-pull-request-lite-pipeline/detail/node-test-pull-request-lite-pipeline/1974/pipeline

[danbev]

CI: https://ci.nodejs.org/job/node-test-pull-request/19592/ ✔️

[addaleax]

Landed in 67f9582

[joyeecheung]

Does this mean we need to treat bindings loaded in workers with care for situations like this? (And this one fails more often because the error map creation is a lot of code when expanded?)

(Also I think we should be able to build that lazily...)
(EDIT: oh no, we also have the constants defined on the binding that are probably relied on by the users...)

[addaleax]

@joyeecheung I think the reason for this is that Map operations call into builtin function, i.e. it is similar to calling JS.

I mean, I would personally prefer to always code more defensively, but I assume people have differing opinions on this…

[joyeecheung]

@addaleax I think it makes sense to be more defensive at least for the binding initializers - if I understand correctly, workers are special because now you have the ability to terminate a thread while it's bootstrapping (whereas previously things were more deterministic on the main thread?)

[addaleax]

@joyeecheung Yes, exactly. I still plan to go through all of our source code and look for problematic cases… which is, unfortunately, a lot of code to go through 😄

[Trott]

Fixed up backport with suggestion from @addaleax and reopened it: #28832