This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch, testsuite] Remove bad test gcc.dg/pr35729.c
On Wed, 2009-01-21 at 21:04 +0100, Richard Guenther wrote:
> Why not restrict the test to a target that is known to work, like for example
> x86_64 or i?86?
>
> Richard.
That seems reasonable. It doesn't fail on my x86_64 run but I do see
failures in other x86_64 runs from gcc-testresults when -fPIC is used.
I don't see any i?86 failures there but I don't see any -fPIC runs with
32 bit x86. I think x86 and nonpic should work. tree-ssa/loop-3.c uses
this setup so I stole that code and comment.
Steve Ellcey
sje@cup.hp.com
$ svn diff gcc.dg/pr35729.c
Index: gcc.dg/pr35729.c
===================================================================
--- gcc.dg/pr35729.c (revision 143537)
+++ gcc.dg/pr35729.c (working copy)
@@ -1,4 +1,7 @@
-/* { dg-do compile } */
+/* Target is restricted to x86 type architectures, so that we may
+ assume something about memory addressing modes. */
+
+/* { dg-do compile { target { { i?86-*-* x86_64-*-* } && nonpic } } }
*/
/* { dg-options "-Os -fdump-rtl-loop2_invariant" } */
const volatile int g_361 = 3L;