Re: [GLIBC][PATCH v2] Remove strdup inlines

[Wilco Dijkstra <Wilco dot Dijkstra at arm dot com>]

Mike Frysinger wrote:
> On 13 Dec 2016 09:24, Wilco Dijkstra wrote:
> > Mike Frysinger wrote:    
> > > we aren't doing this for many other mem/str funcs.  why should we do it
> > > for these two ?  we should be all in, or not do any.  imo, we should just
> > > omit them and be done unless there is strong/compelling evidence to show
> > > otherwise.  if the only point is to support old/uncommon config combos,
> > > then that isn't a great reason imo.
> 
> > A similar redirection is done for several other functions, including mempcpy, stpcpy
> > and bzero. The namespace issue only exists for non-C90 functions that are used
> > inside GLIBC, so a small subset of all supported functions.
>
> i think you're conflating those here.  if you look closely, it's for C++
> code only, and it's because the signature is different (const-vs-non-const
> return).  it's not for the reason you're doing a #define here.

No I'm not talking about the C++ const inlines, I mean redirects to avoid namespace
issues (and to enable GCC to optimize builtins) for non-C90 functions that are used
inside GLIBC. Eg:

#if (!IS_IN (libc) || !defined SHARED) \
  && !defined NO_MEMPCPY_STPCPY_REDIRECT
/* Redirect calls to __builtin_mempcpy and __builtin_stpcpy to call
   __mempcpy and __stpcpy if not inlined.  */
extern __typeof (mempcpy) mempcpy __asm__ ("__mempcpy");
extern __typeof (stpcpy) stpcpy __asm__ ("__stpcpy");
#endif

# ifndef _ISOMAC
#  ifndef index
#   define index(s, c)  (strchr ((s), (c)))
#  endif
#  ifndef rindex
#   define rindex(s, c) (strrchr ((s), (c)))
#  endif
# endif

# ifdef __USE_MISC
#  define strsep(s, reject) __strsep (s, reject)
# endif

#  if defined __USE_XOPEN_EXTENDED || defined __USE_XOPEN2K8
#   define strdup(s) __strdup (s)
#  endif

#  ifdef __USE_XOPEN2K8
#   define strndup(s, n) __strndup (s, n)
#  endif

Removing these causes linknamespace failures.

The question is what is the best approach - copy the magic for mempcpy/stpcpy (which
means you avoid having to do a whole-GLIBC rename and still get compiler builtin 
optimization), use a define that redirects to a builtin (if it exists) or GLIBC internal name?

Note there are also plenty of calls to functions with underscores, eg. __stpcpy,
__mempcpy, __bzero, so we ideally need to either rename these to use the base name
or redirect to the builtin.

Wilco