This is the mail archive of the
mailing list for the GCC project.
Re: Dead code removal before sibcall pass
> > On Wed, May 21, 2003 at 11:14:06PM +0200, Jan Hubicka wrote:
> > > this patch enables dead code removal before sibcall pass making more sibcalls
> > > to be discovered.
> > How often does this happen in real life, as opposed to
> > silly test cases?
> The removal of dead code is pretty common, of course. I didn't made much
> data on how much it affects the amount of found sibcalls but it looks
> easy enought to do and it is always better to kill dead insns in time.
> I will try to bootstrap GCC and count it.
There are no changes in GCC bootstrap times, there are 10 extra sibcalls
in libstdc++ build, so not big deal.
On the other hand I didn't managed to measure any slowdowns.
> > r~