This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH,testsuite] fix darwin-longlong.c for 32-bit simulator targets
- From: Janis Johnson <janis187 at us dot ibm dot com>
- To: Nathan Froyd <froydnj at codesourcery dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Mon, 12 Jan 2009 14:18:48 -0800
- Subject: Re: [PATCH,testsuite] fix darwin-longlong.c for 32-bit simulator targets
- References: <20090112212454.GA28711@codesourcery.com>
- Reply-to: janis187 at us dot ibm dot com
On Mon, 2009-01-12 at 13:24 -0800, Nathan Froyd wrote:
> gcc.target/powerpc/darwin-longlong.c uses SIGILL to determine whether
> the target has 64-bit instruction support. However, newlib lets
> applications that use signal() link successfully, but the registered
> signal handler doesn't actually get invoked. Therefore, it's better to
> use a dg-require-effective-target test rather than a runtime check.
>
> Tested on powerpc-none-eabi. OK to commit?
The support that's being checked is what's provided by -mpowerpc64,
so "powerpc64" would be a better name than "ppc64".
There's a comment in the new check_* proc "Some simulators are known to
not support VMX instructions."; I assume that should be "powerpc64"
instead of "VMX".
Explicitly add the targets from the (removed) xfail as not supporting
powerpc64 instructions.
Support for powerpc64 instructions depends on more than just being able
to execute them so I think this test should have been restricted to
Darwin in the first place, but the patch is OK with these changes.
Janis
> -Nathan
>
> 2009-01-12 Daniel Jacobowitz <dan@codesourcery.com>
>
> * lib/target-supports.exp (check_effective_target_ppc64): New.
> * gcc.target/powerpc/darwin-longlong.c: Explicitly require 64-bit
> instruction support. Do not check for it at runtime.
>