Re: [PATCH] Disable vrp47 for S/390

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Fri, Nov 20, 2009 at 3:29 AM, Andreas Krebbel
<krebbel@linux.vnet.ibm.com> wrote:
> Hi,
>
> the vrp47 testcase currently fails on S/390. ?The ssa code before vrp
> looks different on S/390 compared to x86_64 due to a different value
> returned by BRANCH_COST. ?Branches on S/390 are relatively cheap due
> to a sophisticated branch prediction unit. ?Therefore during
> gimplification fold_truthop (line 5866) uses more branches for
> function h in vrp47.c than the x86_64 variant.
>

> Index: gcc/testsuite/gcc.dg/tree-ssa/vrp47.c
> ===================================================================
> *** gcc/testsuite/gcc.dg/tree-ssa/vrp47.c.orig ?2009-11-09 11:21:04.000000000 +0100
> --- gcc/testsuite/gcc.dg/tree-ssa/vrp47.c ? ? ? 2009-11-19 17:25:35.000000000 +0100
> ***************
> *** 1,6 ****
> ?/* Skip on MIPS, where LOGICAL_OP_NON_SHORT_CIRCUIT inhibits the setcc
> ? ? optimizations that expose the VRP opportunity. ?*/
> ! /* { dg-do compile { target { ! mips*-*-* } } } */
> ?/* { dg-options "-O2 -fdump-tree-vrp -fdump-tree-dom" } */
>
> ?int h(int x, int y)
> --- 1,9 ----
> ?/* Skip on MIPS, where LOGICAL_OP_NON_SHORT_CIRCUIT inhibits the setcc
> ? ? optimizations that expose the VRP opportunity. ?*/
> ! /* Skip on S/390. ?Lower values in BRANCH_COST lead to two conditional
> ! ? ?jumps when evaluating an && condition. ?VRP is not able to optimize
> ! ? ?this. ?*/
> ! /* { dg-do compile { target { ! mips*-*-* ! s390*-*-* } } } */
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

This isn't correct syntax.


H.J.
> ?/* { dg-options "-O2 -fdump-tree-vrp -fdump-tree-dom" } */
>
> ?int h(int x, int y)
>



-- 
H.J.