[PATCH] mm: Suppress mm/memory.o warning on older compilers if !CONFIG_NUMA_BALANCING

[PATCH] mm: Suppress mm/memory.o warning on older compilers if !CONFIG_NUMA_BALANCING

[Mel Gorman]

The kbuild test robot reported the following after the merge of Automatic
NUMA Balancing when cross-compiling for avr32.

mm/memory.c: In function 'do_pmd_numa_page':
mm/memory.c:3593: warning: no return statement in function returning non-void

The code is unreachable but the avr32 cross-compiler was not new enough
to know that. This patch suppresses the warning.

Signed-off-by: Mel Gorman <mgorman@suse.de>
---
 mm/memory.c |    1 +
 1 file changed, 1 insertion(+)

diff --git a/mm/memory.c b/mm/memory.c
index e6a3b93..23f1fdf 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -3590,6 +3590,7 @@ static int do_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
 		     unsigned long addr, pmd_t *pmdp)
 {
 	BUG();
+	return 0;
 }
 #endif /* CONFIG_NUMA_BALANCING */
 

Re: [PATCH] mm: Suppress mm/memory.o warning on older compilers if !CONFIG_NUMA_BALANCING

[Andrew Morton]

On Mon, 17 Dec 2012 11:49:17 +0000
Mel Gorman <mgorman@suse.de> wrote:

> The kbuild test robot reported the following after the merge of Automatic
> NUMA Balancing when cross-compiling for avr32.
> 
> mm/memory.c: In function 'do_pmd_numa_page':
> mm/memory.c:3593: warning: no return statement in function returning non-void
> 
> The code is unreachable but the avr32 cross-compiler was not new enough
> to know that. This patch suppresses the warning.
> 
> Signed-off-by: Mel Gorman <mgorman@suse.de>
> ---
>  mm/memory.c |    1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/mm/memory.c b/mm/memory.c
> index e6a3b93..23f1fdf 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -3590,6 +3590,7 @@ static int do_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
>  		     unsigned long addr, pmd_t *pmdp)
>  {
>  	BUG();
> +	return 0;
>  }
>  #endif /* CONFIG_NUMA_BALANCING */

Odd.  avr32's BUG() includes a call to unreachable(), which should
evaluate to "do { } while (1)".  Can you check that this is working?

Perhaps it _is_ working, but the compiler incorrectly thinks that the
function can return?

Re: [PATCH] mm: Suppress mm/memory.o warning on older compilers if !CONFIG_NUMA_BALANCING

[David Rientjes]

On Mon, 17 Dec 2012, Andrew Morton wrote:

> > The kbuild test robot reported the following after the merge of Automatic
> > NUMA Balancing when cross-compiling for avr32.
> > 
> > mm/memory.c: In function 'do_pmd_numa_page':
> > mm/memory.c:3593: warning: no return statement in function returning non-void
> > 
> > The code is unreachable but the avr32 cross-compiler was not new enough
> > to know that. This patch suppresses the warning.
> > 
> > Signed-off-by: Mel Gorman <mgorman@suse.de>
> > ---
> >  mm/memory.c |    1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/mm/memory.c b/mm/memory.c
> > index e6a3b93..23f1fdf 100644
> > --- a/mm/memory.c
> > +++ b/mm/memory.c
> > @@ -3590,6 +3590,7 @@ static int do_pmd_numa_page(struct mm_struct *mm, struct vm_area_struct *vma,
> >  		     unsigned long addr, pmd_t *pmdp)
> >  {
> >  	BUG();
> > +	return 0;
> >  }
> >  #endif /* CONFIG_NUMA_BALANCING */
> 
> Odd.  avr32's BUG() includes a call to unreachable(), which should
> evaluate to "do { } while (1)".  Can you check that this is working?
> 
> Perhaps it _is_ working, but the compiler incorrectly thinks that the
> function can return?
> 

This isn't the typical "control reaches end of non-void function", the 
warning is merely stating there is no return statement in the function 
which happens to be the case (and it has nothing to do with avr32, it 
will be the same on all archs).  This is one of the last things that gcc 
does after it parses a function declaration and will be emitted with 
-Wreturn-type unless the function in question is main() and it isn't 
marked with __attribute__((noreturn)).  If you're testing this, try making 
the function statically defined and it should show up even with 
do {} while(1).

And for CONFIG_BUG=n this ends up being do {} while (0) which is just a 
no-op and would end up returning that "control reaches end of non-void 
function" warning.