Re: [PATCH] mm/slub: quiet the clang warning with -Wunused-function enabled

[Suren Baghdasaryan <surenb@google.com>]

On Thu, Jul 11, 2024 at 12:43 AM Vlastimil Babka <vbabka@suse.cz> wrote:
>
> On 7/10/24 11:40 PM, Suren Baghdasaryan wrote:
> > On Wed, Jul 10, 2024 at 8:02 AM Nathan Chancellor <nathan@kernel.org> wrote:
> >>
> >> On Wed, Jul 10, 2024 at 04:03:33AM +0100, Matthew Wilcox wrote:
> >> > On Wed, Jul 10, 2024 at 10:54:18AM +0800, sxwjean@me.com wrote:
> >> > > From: Xiongwei Song <xiongwei.song@linux.dev>
> >> > >
> >> > > The only user of prepare_slab_obj_exts_hook() is
> >> > > alloc_tagging_slab_alloc_hook(), which can build with
> >> > > CONFIG_MEM_ALLOC_PROFILING enabled. So, the warning was triggerred
> >> > > when disabling CONFIG_MEM_ALLOC_PROFILING. Let's add "__maybe_unused"
> >> > > for prepare_slab_obj_exts_hook().
> >> >
> >> > Perhaps instead clang can be fixed to match gcc's behaviour?
> >>
> >> Clang only differs from GCC on warning for unused static inline functions in .c
> >> files, not .h files. The kernel already handles this in
> >> include/linux/compiler_types.h but it disables this workaround for W=1 to catch
> >> unused functions like this as a result of commit 6863f5643dd7 ("kbuild: allow
> >> Clang to find unused static inline functions for W=1 build"):
> >>
> >> /*
> >>  * GCC does not warn about unused static inline functions for -Wunused-function.
> >>  * Suppress the warning in clang as well by using __maybe_unused, but enable it
> >>  * for W=1 build. This will allow clang to find unused functions. Remove the
> >>  * __inline_maybe_unused entirely after fixing most of -Wunused-function warnings.
> >>  */
> >> #ifdef KBUILD_EXTRA_WARN1
> >> #define __inline_maybe_unused
> >> #else
> >> #define __inline_maybe_unused __maybe_unused
> >> #endif
> >>
> >> So I don't really think there is much for clang to do here and I think having
> >> the ability to find unused static inline functions in .c files is useful (you
> >> might disagree, perhaps a revert could still be discussed). I guess
> >> IS_ENABLED() can't be used there, so it seems like either taking this patch,
> >> ignoring the warning, or refactoring the code in some other way are the only
> >> options I see.
> >
> > I think this is the consequence of the recent refactoring I've done in
> > https://lore.kernel.org/all/20240704135941.1145038-1-surenb@google.com/.
> > There should be a cleaner way to fix this. I'll post it later today or
> > tomorrow morning.
>
> Yeah looks like the non-empty prepare_slab_obj_exts_hook() could move to the
> #ifdef CONFIG_MEM_ALLOC_PROFILING section above
> alloc_tagging_slab_alloc_hook() and the empty one just removed.

Exactly my plan. I'll post a patch once I reach the office.

>
> > Thanks,
> > Suren.
> >
> >>
> >> > > Reported-by: kernel test robot <lkp@intel.com>
> >> > > Closes: https://lore.kernel.org/oe-kbuild-all/202407050845.zNONqauD-lkp@intel.com/
> >> > > Signed-off-by: Xiongwei Song <xiongwei.song@linux.dev>
> >> > > ---
> >> > >  mm/slub.c | 4 ++--
> >> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> >> > >
> >> > > diff --git a/mm/slub.c b/mm/slub.c
> >> > > index ce39544acf7c..2e26f20759c0 100644
> >> > > --- a/mm/slub.c
> >> > > +++ b/mm/slub.c
> >> > > @@ -2027,7 +2027,7 @@ static inline bool need_slab_obj_ext(void)
> >> > >     return false;
> >> > >  }
> >> > >
> >> > > -static inline struct slabobj_ext *
> >> > > +static inline struct slabobj_ext * __maybe_unused
> >> > >  prepare_slab_obj_exts_hook(struct kmem_cache *s, gfp_t flags, void *p)
> >> > >  {
> >> > >     struct slab *slab;
> >> > > @@ -2068,7 +2068,7 @@ static inline bool need_slab_obj_ext(void)
> >> > >     return false;
> >> > >  }
> >> > >
> >> > > -static inline struct slabobj_ext *
> >> > > +static inline struct slabobj_ext * __maybe_unused
> >> > >  prepare_slab_obj_exts_hook(struct kmem_cache *s, gfp_t flags, void *p)
> >> > >  {
> >> > >     return NULL;
> >> > > --
> >> > > 2.34.1
> >> > >
> >> > >
>