[PATCH] Obvious reload fix

[Alan Modra]

On Sat, Dec 17, 2005 at 01:52:49AM +1030, Alan Modra wrote:
> Tested with a "make quickstrap && make check" with no regressions on
> powerpc64-linux.  Full bootstrap and regression test still in progress.

The full bootstrap failed building libjava due to a bug in
simplify_binary_operation_1.  Paolo, when you moved code over to
simplify-rtx.c, you made some changes and introduced a bug in code I
added here: http://gcc.gnu.org/ml/gcc-patches/2005-09/msg00298.html.

I'm committing the following as obvious.

	* simplify-rtx.c (simplify_binary_operation_1 <IOR>): Correct bug
	introduced 2005-12-16.

Index: gcc/simplify-rtx.c
===================================================================
--- gcc/simplify-rtx.c	(revision 108694)
+++ gcc/simplify-rtx.c	(working copy)
@@ -1972,7 +1972,7 @@ simplify_binary_operation_1 (enum rtx_co
           && (INTVAL (XEXP (SUBREG_REG (opleft), 1)) + INTVAL (XEXP (opright, 1))
               == GET_MODE_BITSIZE (mode)))
         return gen_rtx_ROTATE (mode, XEXP (opright, 0),
-                               XEXP (SUBREG_REG (opright), 1));
+                               XEXP (SUBREG_REG (opleft), 1));
 
       /* If we have (ior (and (X C1) C2)), simplify this by making
 	 C1 as small as possible if C1 actually changes.  */

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre