This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: trivial rs6000 fix
On Mon, Jul 31, 2006 at 09:19:58AM -0700, Stuart Hastings wrote:
>
> On Jul 30, 2006, at 11:16 AM, Mark Mitchell wrote:
>
> >In other words, did this fail without your patch?
>
> Yes, the testcase fails with the current trunk on powerpc-apple-
> darwin, and I expect it would fail on other 64-bit powerpc/rs6000
> targets.
It doesn't fail on powerpc64-linux with "-m64 -fPIC" without the patch.
> I asked Janis if I have the correct DejaGnu voodoo in the testcase;
> haven't heard back yet.
Sorry, I've been out a lot lately.
> Index: gcc.fsf.bigconst/gcc/testsuite/gcc.target/powerpc/20060713-1.c
> ===================================================================
> --- gcc.fsf.bigconst/gcc/testsuite/gcc.target/powerpc/20060713-1.c (revision 0)
> +++ gcc.fsf.bigconst/gcc/testsuite/gcc.target/powerpc/20060713-1.c (revision 0)
> @@ -0,0 +1,9 @@
> +/* { dg-do compile { target { { { powerpc*-*-* rs6000*-*-* } && lp64 } || { powerpc*-apple-* } } } } */
> +/* { dg-options "-fPIC -m64" } */
If you only want to run the test for 64-bit code generation then make
lp64 apply to all targets and leave out the -m64 in the dg-options.
Builds that support -m32/-m64 should run the tests both ways, and the
-m32 test run will skip the test. That is,
/* { dg-do compile { target { lp64 } } } */
/* { dg-options "-fPIC" } */
On the other hand, why bother to restrict it at all, wouldn't the test
be useful for all targets?
Janis