Re: [PATCH v1] rust: Add pr_*_once macros

[Boqun Feng <boqun.feng@gmail.com>]

On Mon, Nov 04, 2024 at 11:08:21PM +0100, jens.korinth@tuta.io wrote:
> Nov 4, 2024, 16:45 by miguel.ojeda.sandonis@gmail.com:
> 
> > On Mon, Nov 4, 2024 at 3:08 AM FUJITA Tomonori
> > <fujita.tomonori@gmail.com> wrote:
> >
> >>
> >> The code to execute only once is very basic code like the kind found
> >> in textbooks? I think that his focus is on optimizing that code. Once
> >> the the basic code is merged, it's easier to send patches to optimize
> >> it.
> >>
> >
> > It doesn't matter that it is "very basic code". This was discussed 2
> > weeks ago in Zulip. Most importantly, it was a new contributor to the
> > kernel that was working on it (as far as I know).
> >
> > The issue is that sending code without any indication about the
> > discussion, without Cc'ing the author, and so on, is the sort of thing
> > that can annoy people. And if they are new contributors, it can drive
> > them away.
> >
> > We need to be careful about this.
> >
> I think you both make good points. Fujita, you're right - it is very
> basic code with little technical complexity, so it may not be worth
> overthinking. At the same time, I really appreciate you including me

I wouldn't call it (adding an extra load() before xchg()) "overthinking"
;-), because we are working in kernel, and we care about performance. We
actually use that pattern everywhere in the kernel. So thank you for
your work on this, the implementation and the benchmarking.

Regards,
Boqun

> in this conversation and acknowledging my contribution, Miguel;
> communities can be quite sensitive, and sometimes, attention to the
> small details really does make a difference. Perhaps we could simply
> add a "Suggested-by" or "Co-developed-by" acknowledgment and move
> forward?
> 
> FWIW, my motivation to look into it was the TODO kernel/error.rs:123.
> Might be a good first place to use the new macros.
> 
> Jens