Re: [GIT PULL] Btrfs fixes for 6.18-rc2

[David Sterba <dsterba@suse.cz>]

On Sun, Oct 19, 2025 at 03:17:32PM -1000, Linus Torvalds wrote:
> On Sun, 19 Oct 2025 at 11:53, Rasmus Villemoes <ravi@prevas.dk> wrote:
> >
> > I think this has come up before [*]. Doesn't -fms-extensions allow one
> > to do
> 
> I clearly have some goldfish genes, because I had completely forgotten
> about that whole similar conversation.
> 
> Yeah, so now we've had at least two use-cases for that thing, although
> from that older discussion we'd apparently need both
> 
>   -fms-extensions
> 
> and
> 
>   -Wno-microsoft-anon-tag
> 
> to also make clang happy about it.
> 
> But yeah, if all versions of gcc and clang that we support do accept
> this thing, maybe we should just bite the bullet and do it, because
> it's just universally useful to be able to define a common helper
> structure and then use it in other structures without naming it. Kind
> of standard "inheritance" syntax, and very useful for that "I have two
> or more parts to this object".
> 
> Want to take that up and see if the btrfs people like the end result?

No objections from me. The other effect of enabling the extensions seems
to be useful wrt the other references in
https://lore.kernel.org/all/20251020142228.1819871-2-linux@rasmusvillemoes.dk/