Re: [GIT PULL] platform-drivers-x86 for v7.2-1

[Linus Torvalds]

On Tue, 23 Jun 2026 at 10:25, Daniel Lezcano
<daniel.lezcano@xxxxxxxxxxxxxxxx> wrote:
>
> Yes, sorry for that. I'll take care of not reusing the same function
> name in the future

Note that reusing a name *eventually* is fine - it's not like people
need to try to remember that some name was used in the distant past
and never ever use that name again.

Yes, even that can then obviously cause confusion for stable backports
if somebody is really unlucky and ends up using a function that has
changed meaning over time, but at some point it's just bad luck and
re-using a good name is fine.

So it's just the "don't quickly re-use a name with entirely different
semantics" because that really does end up very confusing.

I entirely understand why the resolution in linux-next ended up being
what it was: Mark clearly just assumed it was the normal kind of
"calling convention changed with new arguments".

Because that "extend the calling convention" thing is pretty common,
and that was my first reaction to the conflict too.

It was only when I looked at it a second time and went "hey, wait a
minute..." that I started looking through the history of that
function. I could _very_ easily have missed it entirely.

Now, things could be even worse: re-using the name for something very
different, and then having the same calling conventions as the old
name for completely new semantics.

Then it would have resulted in a completely silent and very subtle
semantic merge conflict.

So symbol name re-use just needs some care, and some distance in time.
It's not a complete no-no, it just ends up easily causing mistakes.

             Linus