Re: [PATCH] net/sunrpc/svc.c: fix sparse warning

["J. Bruce Fields" <bfields@fieldses.org>]

On Thu, May 14, 2009 at 11:58:37AM -0400, H Hartley Sweeten wrote:
> On Monday, May 11, 2009 4:06 PM, J. Bruce Fields wrote:
> > On Thu, Apr 16, 2009 at 03:15:17PM -0400, H Hartley Sweeten wrote:
> >> On Thursday, April 16, 2009 12:06 PM, J. Bruce Fields wrote:
> >>> On Wed, Apr 15, 2009 at 03:46:13PM -0400, H Hartley Sweeten wrote:
> >>>> Fix sparse warning in net/sunrpc/svc.c.
> >>>> 
> >>>> 	warning: symbol 'node' shadows an earlier one
> >>>
> >>> What's the other symbol?
> >>
> >> Sorry. Left that out...
> >> 
> >> include/linux/nodemask.h
> >> 
> >> #define any_online_node(mask)			\
> >> ({						\
> >> 	int node;				\
> >> 	for_each_node_mask(node, (mask))	\
> >> 		if (node_online(node))		\
> >> 			break;			\
> >> 	node;					\
> >> })
> >> 
> >> arch/powerpc/mm/numa.c is the only other user of that macro. In that
> >> file the local variable is called nid, hence the name change in this
> >> patch.
> >
> > Stupid question (and sorry for the delay): any reason that macro
> > couldn't just be a static inline function?
> >
> > I'm sort of resistant to the idea that the caller should have to care
> > what local variable names the implementation uses.
> 
> That would probably be cleaner and it would prevent the same warning
> from possibly occurring in the future.  I just assumed it was originally
> made a macro for a good reason.
> 
> Should I submit an updated patch changing the macro into an static
> inline function?

I'd be for it.  If nothing else that'll probably be the fastest way to
find out if there was a good reason!

--b.