gh-127481: Add EPOLLWAKEUP to the select module

[rruuaanng]

• Issue: Add EPOLLWAKEUP to the select module #127481

---

📚 Documentation preview 📚: https://cpython-previews--127482.org.readthedocs.build/

[picnixz]

The EPOLL_URING_WAKE is documented as being internal only. Similarly, EPOLLWAKEUP has some constraints:

/*
 * Request the handling of system wakeup events so as to prevent system suspends
 * from happening while those events are being processed.
 *
 * Assuming neither EPOLLET nor EPOLLONESHOT is set, system suspends will not be
 * re-allowed until epoll_wait is called again after consuming the wakeup
 * event(s).
 *
 * Requires CAP_BLOCK_SUSPEND
 */

I'm not sure we should add EPOLL_URING_WAKE. As for EPOLLWAKEUP, I think a .. versionadded:: next entry could be good (it also requires Linux Kernel 4.5 or later so such mention is worth adding).

[rruuaanng]

Yes, it's a restricted function, but it does not prevent us from providing it to users who turn it on.

Edit

I think a .. versionadded:: next entry could be good (it also requires Linux Kernel 4.5 or later so such mention is worth adding).

Good suggestion. I'll add the version number that appeared for the first time tomorrow. It's too late today.

But for versionadded, it may not be necessary. If I remember correctly, it seems that there is no this entry in the document.
(I think.)

[picnixz]

But for versionadded, it may not be necessary. If I remember correctly, it seems that there is no this entry in the document.

There is something just after the table... It says exactly what I expect the entry to say.

[rruuaanng]

But for versionadded, it may not be necessary. If I remember correctly, it seems that there is no this entry in the document.

There is something just after the table... It says exactly what I expect the entry to say.

Ehm... I saw it :(, I'll add it.

[ZeroIntensity]

CI failures look unrelated. I'll investigate.

[ZeroIntensity]

So, based on the 66 return code that looks like an actual data race. But, I wasn't able to find anything recent related to that test, so it must be exceptionally rare. Re-running jobs seemed to do the trick, someone @ me if you notice it on another PR.

[rruuaanng]

Thank you for your work. If there is any problem, I'll contact you as soon as problem :)

[encukou]

Thank you!

[rruuaanng]

@encukou Thank you for your merge.