This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [patches] Re: more robust purge_dead_edges


> On Tue, Jul 24, 2001 at 11:45:11PM +0200, Jan Hubicka wrote:
> > +       /* We do care only about conditional jumps and simplejumps.  */
> > +       if (!any_condjump_p (insn)
> > + 	  && !simplejump_p (insn))
> > + 	return;
> 
> How about a conditional return being simplified to an
> unconditional return?  This wouldn't seem to support that.
It does get to the earlier path:

+       /* Redistribute probabilities.  */
+       if (!bb->succ->succ_next)
+ 	{
+ 	  bb->succ->probability = REG_BR_PROB_BASE;
+ 	  bb->succ->count = bb->count;
+         }
+       else
+ 	{
+ 	  note = find_reg_note (insn, REG_BR_PROB, NULL);
+ 	  /* We should have prediction on each conditional jump.  */
+ 	  if (!note)
+ 	    abort ();


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]