Here's a simple test case that fails on the main branch.

https://github.com/sourcejedi/cpython/commit/50184ea3b354fd775866d036ccee058ec6734927

> the patch assumes that python signal handler will be called *before* reading from self-pipe. 
> Otherwise, a signal callback will be postponed up to next writing to the pipe, which looks like a hidden bug.

I think it will be?  I can imagine it being a concern for future changes though.  I thought it could break if anyone takes seriously the comment "XXX Signals should be recorded per thread, now we have thread state." :-).

But if we're not confident, surely all we have to do is make sure to call _write_to_self() at the end of the python signal handler (_UnixSelectorEventLoop._sig_handler()).

It seems a pity not to have example code we can point to, that handles signals correctly.  Even if it's too tedious to explain the correct way in the official docs.