Re: Should we optimize memset to bzero?

[Jakub Jelinek <jakub at redhat dot com>]

On Fri, Nov 02, 2001 at 08:44:06AM -0800, H . J . Lu wrote:
> On Fri, Nov 02, 2001 at 05:38:00PM +0100, Jakub Jelinek wrote:
> > On Fri, Nov 02, 2001 at 08:28:49AM -0800, H . J . Lu wrote:
> > > Should we optimize memset to bzero? Like
> > > 
> > > #define memset(s, c, n) \
> > > (__extension__ ((__builtin_constant_p (c) && (c) == 0) \
> > > 		? __bzero ((s), (n)) : __memset ((s), (c), (n)))
> > 
> > And we don't? Look at bits/string2.h...
> 
> We do? Try this
> 
> --s.c--
> #include <string.h>
> 
> void
> foo (char *x, int l)
> {
>   memset (x, 0, l);
> }
> ---
> 
> # gcc -O2 -S s.c
> 
> BTW, in many cases, gcc won't inline those string/memory functions.

Well, the thing is that glibc further optimizes __bzero(s,n) with
__builtin_memset(s,0,n); which often generates better code.
But, teaching gcc that it can use __bzero is difficult, since __bzero
is not part of any standard.

	Jakub