Re: PR 22051

["Joseph S. Myers" <joseph at codesourcery dot com>]

On Sat, 25 Jun 2005, Jeffrey A Law wrote:

> > Mainline testing on hppa2.0w-hp-hpux11.23 shows a new FAIL
> > 
> > FAIL: gcc.dg/tree-ssa/20030807-2.c scan-tree-dump-times if  0
> > 
> > but this may not be indicative of a problem with the patch 
> Is indicative of a problem with the patch.  That test is verifying
> that we are eliminating a cast from one pointer type to another
> when the resulting pointer is used in a comparison against null.

I can confirm we have the same new failure with the patch on 4.0 branch as 
well as on mainline.

Tree dump, 4.0 branch, before the patch:


;; Function oof (oof)

oof ()
{
  int * live;
  int live_head;

<bb 0>:
  bitmap_clear (&live_head);
  return;

}



;; Function foo (foo)

foo (n)
{
  int * space;
  void * D.1130;
  long unsigned int n___2;

<bb 0>:
  n___2_2 = (long unsigned int) n_1;
  D.1130_3 = __builtin_alloca (n___2_2);
  space_4 = (int *) D.1130_3;
  bar (space_4);
  return;

}

Diff between tree dumps before and after:

--- 20030807-2.c.t57.dom3.old   2005-06-25 09:37:33.000000000 -0700
+++ 20030807-2.c.t57.dom3       2005-06-25 09:38:18.000000000 -0700
@@ -26,6 +26,12 @@
   n___2_2 = (long unsigned int) n_1;
   D.1130_3 = __builtin_alloca (n___2_2);
   space_4 = (int *) D.1130_3;
+  if (space_4 == 0B) goto <L0>; else goto <L1>;
+
+<L0>:;
+  abort ();
+
+<L1>:;
   bar (space_4);
   return;
 

-- 
Joseph S. Myers               http://www.srcf.ucam.org/~jsm28/gcc/
    jsm@polyomino.org.uk (personal mail)
    joseph@codesourcery.com (CodeSourcery mail)
    jsm28@gcc.gnu.org (Bugzilla assignments and CCs)