Re: [RFC] link_path_walk cleanup

[Denis Vlasenko]

On 26 April 2002 15:26, Alexander Viro wrote:
> On Fri, 26 Apr 2002, Andrew Morton wrote:
> > Maneesh Soni wrote:
> > >..
> > > +static inline int walk_one(struct nameidata *nd)
> >
> > This function is hundreds and hundreds of bytes of code.  It has
> > three call sites.  Making it an inline is very inefficient!
>
> Unfortunately, this is a very special case.  This sucker is involved
> in mutual recursion and extra frame on stack will be nasty for, say it,
> sparc.  Normally I would agree that something of that kind should not be
> inlined, but...

Maybe add a comment? Or this question will emerge on lkml from time to time.
--
vda