Re: [PATCH v8 0/7] Add pidfd and cgroupv2 support for process creation

[Florian Weimer <fweimer@redhat.com>]

* Luca Boccassi:

> These are real race conditions, that cannot be solved otherwise,
> characterizing them as 'programming errors' is very misleading and
> wrong.
>
> We very much need both of those interfaces in systemd, and fully intend
> to use them as soon as they are available. We are slowly moving towards
> using pidfds everywhere to be able to do end-to-end race-free process
> tracking and management, and these are fundamental pieces for this
> effort. From what I can read the GNOME developers feel the same way,
> and I wouldn't be surprised if QT followed suit too given what you
> mentioned in the cover letter.
>
> Surely implementing useful, core functionality for the direct and
> immediate benefit of 3 major Linux projects is a reason as solid as you
> could ever find to add a new interface.

I see value in adding fork support, too.

The fundamental issue with the fork part is that it's not future-proof
at all.  The programming model is completely different from the kernel
interface.

If I start a discussion about API alternatives, who should I Cc: except
you and Rich?

Thanks,
Florian