Re: [PATCH] tools: fix -Wunused-result in linux.c

[Shuah Khan]

On 10/21/24 10:10, David Laight wrote:
> From: Liam R. Howlett
> > Sent: 15 October 2024 02:11
> >
> > * Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx> [241011 18:52]:
> > > Fix the following -Wunused-result warnings on posix_memalign()
> > > return values and add error handling.
> > >
> > > ./shared/linux.c:100:25: warning: ignoring return value of ‘posix_memalign’ declared with attribute
> > ‘warn_unused_result’ [-Wunused-result]
> > >    100 |          posix_memalign(&p, cachep->align, cachep->size);
> > >        |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > ../shared/linux.c: In function ‘kmem_cache_alloc_bulk’:
> > > ../shared/linux.c:198:33: warning: ignoring return value of ‘posix_memalign’ declared with attribute
> > ‘warn_unused_result’ [-Wunused-result]
> > >    198 |          posix_memalign(&p[i], cachep->align,
> > >        |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > >    199 |                                cachep->size);
> > >        |                                ~~~~~~~~~~~~~
> > >
> > > Signed-off-by: Shuah Khan <skhan@xxxxxxxxxxxxxxxxxxx>
> >
> > Reviewed-by: Liam R. Howlett <Liam.Howlett@xxxxxxxxxx>
> >
> > > ---
> > >   tools/testing/shared/linux.c | 14 +++++++++-----
> > >   1 file changed, 9 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/tools/testing/shared/linux.c b/tools/testing/shared/linux.c
> > > index 17263696b5d8..66dbb362385f 100644
> > > --- a/tools/testing/shared/linux.c
> > > +++ b/tools/testing/shared/linux.c
> > > @@ -96,10 +96,13 @@ void *kmem_cache_alloc_lru(struct kmem_cache *cachep, struct list_lru *lru,
> > >   		p = node;
> > >   	} else {
> > >   		pthread_mutex_unlock(&cachep->lock);
> > > -		if (cachep->align)
> > > -			posix_memalign(&p, cachep->align, cachep->size);
> > > -		else
> > > +		if (cachep->align) {
> > > +			if (posix_memalign(&p, cachep->align, cachep->size) < 0)
> > > +				return NULL;
> > > +		} else {
> > >   			p = malloc(cachep->size);
> > > +		}
> > > +
>
> You really ought to be checking malloc() as well.
> Perhaps:
> 		if (...) {
> 			if (posix_memalign(...) < 0)
> 				p = NULL;
> 		} else {
> 			p = malloc(...);
> 		}
> 		if (!p)
> 			return NULL;
>
> Or just use a hack to get the compiler to STFU :-)

Yes you are right. I will spin another version to cover
the malloc and I just noticed another posix_memalign()
with -Wunused-result that compiler didn't flag in
kmem_cache_alloc_bulk() to fix as well in the next version.

thanks,
-- Shuah