add lowlevel thread support

[rainers]

another extract of #2514

[dlang-bot]

Thanks for your pull request, @rainers!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub fetch digger
dub run digger -- build "master + druntime#2519"

[CyberShadow]

Could the high-level thread code be refactored on top of the low-level code added here?

[rainers]

Could the high-level thread code be refactored on top of the low-level code added here?

Possible, but with a couple of complications, because some high level variables are changed inbetween (e.g. m_hndl, m_isRunning). That would probably need a greater refactoring.

[CyberShadow]

If the symbols added here are to be public, then I think that would make this a situation that @andralex asked to be notified about.

[rainers]

I wouldn't mind keeping the symbols "private", i.e. undocumented, but I want to call them from the GC.

[jacob-carlborg]

I wouldn't mind keeping the symbols "private", i.e. undocumented, but I want to call them from the GC.

The GC is not in the core top level package?

[rainers]

The GC is not in the core top level package?

No, it has its own top level package gc.

[thewilsonator]

This needs another patch to weka-io/mecca

[thewilsonator]

That patch for weka is blocked on https://issues.dlang.org/show_bug.cgi?id=19761

[thewilsonator]

Why does CircleCI fail to install DMD?

[thewilsonator]

This otherwise looks go to me but CircleCI is required.

[ibuclaw]

Note: this is alignment is insufficient for targets with strict alignment requirements.

[rainers]

Yes, will need to be updated in lock-step with Thread._locks.