libffi alpha--freebsd* configuration patch

Fri Mar 22 13:57:00 GMT 2002

In article <87r8mcbhhb.fsf@creche.redhat.com> you write:
>>>>>> "Loren" == Loren James Rittle <rittle@latour.rsch.comm.mot.com> writes:

> Loren> I can report that libffi works for alpha*-*-freebsd* when the obvious
> Loren> enablement patch is applied to libffi/configure.in . ./ffitest
> Loren> reports: ``[...] Looks good. No surprises.'' Applied ``as obvious''
> Loren> to the mainline after it checked as expected.

> Do we want this on the branch as well?

[Since this is a general port status report, I have redirected the
 answer to the general java list.]
I assume so but until I get all of libjava building on the platform,
the point is somewhat moot (I promise to ask permission to move all
required patches in sequence once it works on mainline and I test on
branch - I *really* want to get it enabled by default by the 3.1
release). I now have a recent enough gdb build for alpha*-*-freebsd*
thus can debug the last known (to me) showstopping issue:
../../gcc/gcjh -classpath '' -bootclasspath . java/lang/Double
gmake[1]: *** [java/lang/Double.h] Floating point exception (core dumped)
By default, I think this alpha port throws an exception on overflow
whereas i386-*-freebsd* does not.
However, I do also see that the number of runtime failures has shot
way up on i386-*-freebsd* recently.
Results for 3.1 20020122 (experimental) testsuite on
i386-unknown-freebsd5.0 were (freebsd4.5 was similar into February):
# of unexpected failures 36
[http://gcc.gnu.org/ml/gcc-testresults/2002-01/msg00508.html]
[http://gcc.gnu.org/ml/gcc-testresults/2002-02/msg00129.html]
Today it is now over 100.
[http://gcc.gnu.org/ml/gcc-testresults/2002-03/msg00284.html]
I intend to slug away on it today and as much time as I can next week.
Regards,
Loren

libffi alpha*-*-freebsd* configuration patch

libffi alpha--freebsd* configuration patch