This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: ifcvt: cond_exec-like behavior on cc0 machines
On Jul 21, 2003, Richard Henderson <rth@redhat.com> wrote:
> (2) This is false. The cc0 setter must *must* MUST be the previous
> instruction.
Ok, let's assume that it is. Two conditions that I remember that I
think may have caused me grief while investigating the use of
cond_exec:
(set (cc0) (reg:M)) ;; tstM
(set (cc0) (compare (reg:M) (reg_or_const:M))) ;; cmpM
IIRC, get_condition() may have failed in a case that had one of
these setting (cc0) before the branch. I'll have to go back and try
to recreate the condition, though.
> From the point of view of flow, both *ought* to work.
I'd got the impression that cond_exec was fundamentally incompatible
with cc0. If this is not the case, I'll be more than happy to try to
fix the problems.
> But I'm as yet unconvinced that ifcvt needs to add support for the
> later if its effectively identical to the former.
Agreed.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist Professional serial bug killer