Re: [PATCH 04/13] mm: Fix pmd_read_atomic()

[Linus Torvalds]

On Tue, Nov 1, 2022 at 5:42 AM Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> So I've stuck the below on. There's a *TON* more to convert and I'm not
> going to be doing that just now (seems like a clever enough script
> should be able to), but this gets rid of the new one I introduced.

Thanks.

And no, I don't think the churn of converting old cases is worth it. I
just want to discourage *more* of this.

Using the same name really helps when you do a "git grep" for a
symbol, the whole '#ifndef' patterns for alternate architecture
definitions shows up really clearly.

So that - together with not having the possibility of mixing up names
- is the main reason I don't like the ARCH_HAS_XYZ pattern, and much
prefer just using the name of whichever function gets an architecture
override.

                 Linus

PS. I'd love to get an ack/nak on the "mm: delay rmap removal until
after TLB flush" thing.