[Bug middle-end/48124] [4.4/4.5/4.6/4.7 Regression] likely wrong code bug

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48124

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|52080                       |

--- Comment #9 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-02-01 12:34:55 UTC ---
Simpler patch I am going to test.  Let's hope the wreckage adjust_address
does to the to_rtx MEM (apart from setting its mode) is harmless.

Index: expr.c
===================================================================
--- expr.c      (revision 183791)
+++ expr.c      (working copy)
@@ -4705,6 +4705,21 @@ expand_assignment (tree to, tree from, b
            to_rtx = adjust_address (to_rtx, mode1, 0);
          else if (GET_MODE (to_rtx) == VOIDmode)
            to_rtx = adjust_address (to_rtx, BLKmode, 0);
+         /* If the alignment of tem is larger than its size and we
+            are performing a bitfield access limit the mode we use
+            for the access to make sure we do not access the decl
+            beyond its end.  See PR48124.  */
+         else if (GET_MODE (to_rtx) == BLKmode
+                  && mode1 == VOIDmode
+                  && DECL_P (tem)
+                  && TREE_CODE (DECL_SIZE (tem)) == INTEGER_CST
+                  && TREE_INT_CST_LOW (DECL_SIZE (tem)) % DECL_ALIGN (tem))
+           {
+             mode1 = mode_for_size (TREE_INT_CST_LOW (DECL_SIZE (tem))
+                                    % DECL_ALIGN (tem),
+                                    MODE_INT, 0);
+             to_rtx = adjust_address (to_rtx, mode1, 0);
+           }
        }

       if (offset != 0)