Re: [PATCH net-next V3 04/11] net/mlx5: fs, add mlx5_fs_pool API

[Jakub Kicinski <kuba@kernel.org>]

On Thu, 19 Dec 2024 14:30:41 +0200 Moshe Shemesh wrote:
> > Locally (say two lines above) your label name is obvious.
> > But please imagine it in the context of whole function, it is much
> > better to name labels after what they jump to (instead of what they
> > jump from). It is not only easier to reason about, but also more
> > future proof. I think Simon would agree.
> > I'm fine with keeping existing code as-is, but for new code, it's
> > always better to write it up to the best practices known.
>
> I tend to name labels according to what they jump from. Though if I see 
> on same function labels are used the other way I try to be consistent 
> with current code.
> I think there are pros and cons for both ways and both ways are used.
> I can change here, but is that kernel or netdev consensus ?

Yes, there's a consensus now. But I think since all mlx* code uses
the "jump source" naming mixing the two could lead to confusion for
your internal developers, and bugs. So I'd say up to you :(
-- 
Since Przemek found a real bug elsewhere:
pw-bot: cr