Andy Pan

Ping @madolson

> @panjf2000 Ok. I just can't convince myself this is a good change to take right now. We know we don't handle edge cases well, so I'm not able to...

I've finally discovered the root cause that had been failing the tests on macOS before, `uv__async_send()` might get called before `uv__io_poll()` with multi-threads, which resulted in an `ENOENT` returned from...

There might be some code conflicts with #4400.

Some references of `EVFILT_USER` on various platforms: - [FreeBSD kqueue](https://man.freebsd.org/cgi/man.cgi?query=kqueue) - [NetBSD kqueue](https://man.netbsd.org/NetBSD-10.0/kqueue.2) - [DragonFlyBSD annotated tag v4.0.0rc created](https://lists.dragonflybsd.org/pipermail/commits/2014-October/343769.html) - [macOS swift-nio](https://github.com/apple/swift-nio/blob/196ceab4b21ba5b84eea522f9829db31b7b91f19/Sources/NIOPosix/SelectorKqueue.swift#L167-L182) - [XNU kernel](https://github.com/apple-oss-distributions/xnu/blob/94d3b452840153a99b38a3a9659680b2a006908e/tests/kqueue_in_select.c#L42-L50)

> Okay, it wasn't clear that EVFILT_USER was available on XNU. Makes more sense now that is clear it is available in the XNU sources, just not documented. Could you...

Any new thoughts on this? @bnoordhuis @vtjnash @saghul

Ping @bnoordhuis @vtjnash

Ping @libuv/collaborators

I see @saghul gave his approval on this a long time ago but I guess that will not suffice to merge this PR. Since @bnoordhuis @vtjnash seemed to have got...