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]
Other format: [Raw text]

Re: [tree-ssa] must-alias fixes...


> On Fri, Nov 21, 2003 at 01:56:48PM +0100, Jan Hubicka wrote:
> > Does your changes imply that the NOPs never get inside PLUS?
> 
> My changes don't imply that.  It never should have happened
> before my changes either.  
OK, I will remove the last NOP skipping code from the patch, re-test and
commit if passes.

Would be the attached change into tree-simple OK (assuming it passes
testing with my verify_stmt patch in)?  I think we should
ensure the canonical form of addresses.

Index: tree-simple.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/Attic/tree-simple.c,v
retrieving revision 1.1.4.61
diff -c -3 -p -r1.1.4.61 tree-simple.c
*** tree-simple.c	17 Nov 2003 23:18:13 -0000	1.1.4.61
--- tree-simple.c	21 Nov 2003 13:33:17 -0000
*************** is_gimple_min_invariant (tree t)
*** 294,301 ****
  
      case PLUS_EXPR:
        return (TREE_INVARIANT (t)
! 	      && is_gimple_min_invariant (TREE_OPERAND (t, 0))
! 	      && is_gimple_min_invariant (TREE_OPERAND (t, 1)));
  
      case INTEGER_CST:
      case REAL_CST:
--- 294,301 ----
  
      case PLUS_EXPR:
        return (TREE_INVARIANT (t)
! 	      && TREE_CODE (TREE_OPERAND (t, 0)) == ADDR_EXPR
! 	      && TREE_CODE (TREE_OPERAND (t, 1)) == INTEGER_CST);
  
      case INTEGER_CST:
      case REAL_CST:


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