Re: [PATCH 23/25] ALSA: hda/core: Use guard() for spinlocks

[Takashi Iwai <tiwai@suse.de>]

On Wed, 13 Aug 2025 16:10:05 +0200,
Andy Shevchenko wrote:
> 
> On Wed, Aug 13, 2025 at 04:07:22PM +0200, Andy Shevchenko wrote:
> > On Mon, Aug 11, 2025 at 12:07:57PM +0200, Takashi Iwai wrote:
> > > Replace the manual spin lock/unlock pairs with guard() for code
> > > simplification.
> > > 
> > > Only code refactoring, and no behavior change.
> > 
> > Wishful thinking :-)
> > (At least) this broke my builds
> > 
> > /sound/hda/core/stream.c:954:2: error: label followed by a declaration is a C23 extension [-Werror,-Wc23-extensions]
> >   954 |         guard(spinlock_irq)(&bus->reg_lock);
> >       |         ^
> > /include/linux/cleanup.h:405:2: note: expanded from macro 'guard'
> >   405 |         CLASS(_name, __UNIQUE_ID(guard))
> >       |         ^
> > /include/linux/cleanup.h:290:2: note: expanded from macro 'CLASS'
> >   290 |         class_##_name##_t var __cleanup(class_##_name##_destructor) =   \
> >       |         ^
> > <scratch space>:36:1: note: expanded from here
> >    36 | class_spinlock_irq_t
> >       | ^
> > 1 error generated.
> > 
> > `make W=1`, clang-19.
> 
> Need to add this is with patches applied from
> https://lore.kernel.org/lkml/20250804220955.1453135-1-dan.j.williams@intel.com/
> as otherwise it doesn't compile differently (in other places).

Thanks for the info.
OK, it seems that the patches need some workarounds for clang.
Will drop for now, then.


Takashi