> You are right, the issue is "theoretical", I didn't see problems with current implementation.
> The self-pipe buffer is pretty big, a chance to reach its limit is relatively low.

Actually, I disagree.  The problem manifests itself easily when you orchestrate a big number of subprocesses (think 1000s).

I've implemented a different strategy for signals handling in uvloop [1] and I think we can easily employ similar approach in asyncio.

Please keep this open. 

[1] https://github.com/MagicStack/uvloop/commit/6e03e513625eb544a5f60c9bf9922019fc3a0b29