This is the mail archive of the
libffi-discuss@sourceware.org
mailing list for the libffi project.
Re: how to run the testsuite
- From: Sebastian Reitenbach <sebastia at l00-bugdead-prods dot de>
- To: sebastia at l00-bugdead-prods dot de, andreast-list at fgznet dot ch
- Cc: libffi-discuss at sourceware dot org
- Date: Thu, 11 Dec 2008 12:10:23 +0100
- Subject: Re: how to run the testsuite
- Reply-to: Sebastian Reitenbach <sebastia at l00-bugdead-prods dot de>
Hi,
Andreas Tobler <andreast-list@fgznet.ch> wrote:
> Sebastian Reitenbach wrote:
>
> > that worked so, far, but some unexpected failures happened:
> > === libffi Summary ===
> >
> > # of expected passes 1364
> > # of unexpected failures 9
> > # of unsupported tests 15
> > *** Error code 1
> >
> > this is on OpenBSD 4.3 sparc64, libffi-3.0.7, gcc-3.3.5 installed, the
logfile
> > is appended.
> >
> > sorry, forgot to append the logfile, meanwhile the sparc is also done
with the
> > test. the sparc runs OpenBSD 4.4, gcc-2.95.3, libffi-3.0.7.
>
> Ok, as expected, the 64-bit port fails due to a base compiler which is
> too old. The one failure I mentioned open, you can ignore. This is just
> a warning.
>
> The 32-bit port is much more worse, I didn't know that there are
> gcc-2.95 around. No chance with such an old compiler. Sorry.
>
> I think the last GPLv2 compiler is 4.2.1, this one should be a candidate
> for OpenBSD as well. I have at least on FreeBSD good results on powerpc
> and amd64. The failures in the log of the 64-bit ports are gone with a
> recent compiler, with recent I mean one of the 4.x series.
>
> I bet the 32-bit port will behave equally with a recent compiler.
>
> Well, there is one minor issue which might have to be fixed in the
> libffi package for sparc, 32-bit.
>
> 2008-11-21 Eric Botcazou <ebotcazou@adacore.com>
>
> * src/sparc/ffi.c (ffi_prep_cif_machdep): Add support for
> signed/unsigned int8/16 return values.
> * src/sparc/v8.S (ffi_call_v8): Likewise.
> (ffi_closure_v8): Likewise.
>
> This one has to be merged towards libffi on sourceware. I'll ping AG for
> this.
>
> Andreas
>
>
I got gcc-4.2 installed on the sparc64. The configure and make seemed to
have used the environment variables. CC=egcc CXX=eg++ CPP=ecpp, but when I
run "make check", with or without the environment variables, then I see in
the log, it is still using gcc instead of egcc. Is it me doing sth. wrong?
regards
Sebastian