This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [testsuite, ada] Check for .cfi_personality support
Andrew Pinski <pinskia@gmail.com> writes:
> What I am trying to say is that sorry message means not using
> cfi_personality has not been implemented yet. Which means these tests are
> expected to fail. I think it is better to have an xfail here rather than
> unsupported as it would be easier to spot when multiple personality
> functions are supported without cfi_personality support.
Again: how would you do the XFAILing? Still with the new
cfi_personality (or similar) effective-target keyword? Doing this
target-based cannot work: Solaris with as vs. old gas vs. new gas: the
first two don't work, the last does. While this would turn into an
XPASS when multi-personality support were implemented, this doesn't
solve the missing executable problem, so we have testsuite noise for a
known problem.
I'd rather handle this as currently proposed and file a PR for the
multi-personality support with an explicit reference to
check_effective_target_cfi_personality which needs to be removed or
updated once the implementation is done.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University