From patchwork Wed Sep 29 01:36:06 2010
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: canonical condition branch
Date: Tue, 28 Sep 2010 15:36:06 -0000
From: Mike Stump <mikestump@comcast.net>
X-Patchwork-Id: 66034
Message-Id: <93FB9EB0-77DA-4DF2-B25F-2CE73CE30D8E@comcast.net>
To: GCC Patches <gcc-patches@gcc.gnu.org>

I have a port with a pattern like:

(define_insn "*branch_if_bit_set"
  [(set (pc)
        (if_then_else
         (zero_extract:DI (match_operand:DI 0 "register_operand" "r")
                          (const_int 1)
                          (match_operand 1 "immediate" ""))
         (label_ref (match_operand 2 "" ""))
         (pc)))]
  ""
  "branch_if_bit_set\t%0,%1,%2")

I used that pattern as that is what I saw combine looking for and failing to find...  but, in rtlanal.c (canonicalize_condition), we have:

which drops op2 because zero_extract is a 3 operand code, not a 2 operand code.  So, then question is, should we add the above patch to avoid building bad rtl code?


Index: gcc/gcc/rtlanal.c
===================================================================
--- gcc/gcc/rtlanal.c	(revision 850)
+++ gcc/gcc/rtlanal.c	(working copy)
@@ -4933,6 +4933,9 @@
   if (CC0_P (op0))
     return 0;
 
+  if (GET_RTX_LENGTH (code) != 2)
+    return 0;
+
   return gen_rtx_fmt_ee (code, VOIDmode, op0, op1);
 }