Re: [PATCH v3 3/9] s390: Refactor ifunc resolvers due to false debuginfo.

[Stefan Liebler <stli at linux dot vnet dot ibm dot com>]

On 08/30/2016 02:31 PM, Florian Weimer wrote:
> On 08/24/2016 04:04 PM, Stefan Liebler wrote:
> >  #if defined SHARED && IS_IN (libc)
> > +# define mempcpy __redirect_mempcpy
> > +# define __mempcpy __redirect___mempcpy
> > +/* Omit the mempcpy inline definitions because it would redefine
> > mempcpy.  */
> > +# define _HAVE_STRING_ARCH_mempcpy 1
> > +# include <string.h>
> > +# undef mempcpy
> > +# undef __mempcpy
> >  # include <ifunc-resolve.h>
>
> The long-term maintainability of such #define trickery still concerns
> me.  What would be needed to avoid that?
I think we need a possibility to omit the hidden_proto definitions for 
the ifunc'ed symbols when using libc_ifunc macro. Then we wouldn't need 
the redirection and could use the libc_ifunc macro instead of 
libc_ifunc_redirected.
Only thoughts. Not tested if there are other pitfalls.
> (This obviously shouldn't be part of this patch sequence, I'm just
> curious.)
>
> Thanks,
> Florian