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]

fix expand_vec_cond_expr


I think this is a pure think-o.  I can't think of a reason this
was intended to be EXPAND_STACK_PARM (1).  Indeed, I have an ICE
demonstratable on autovect-branch that suggests it isn't.

Applied to mainline, 4.1, and autovect.


r~


        * optabs.c (expand_vec_cond_expr): Use EXPAND_NORMAL.

=== optabs.c
==================================================================
--- optabs.c	(revision 107842)
+++ optabs.c	(local)
@@ -5727,13 +5727,13 @@
   cc_op1 = XEXP (comparison, 1);
   /* Expand both operands and force them in reg, if required.  */
   rtx_op1 = expand_expr (TREE_OPERAND (vec_cond_expr, 1),
-			 NULL_RTX, VOIDmode, 1);
+			 NULL_RTX, VOIDmode, EXPAND_NORMAL);
   if (!insn_data[icode].operand[1].predicate (rtx_op1, mode)
       && mode != VOIDmode)
     rtx_op1 = force_reg (mode, rtx_op1);
 
   rtx_op2 = expand_expr (TREE_OPERAND (vec_cond_expr, 2),
-			 NULL_RTX, VOIDmode, 1);
+			 NULL_RTX, VOIDmode, EXPAND_NORMAL);
   if (!insn_data[icode].operand[2].predicate (rtx_op2, mode)
       && mode != VOIDmode)
     rtx_op2 = force_reg (mode, rtx_op2);


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