libffi alpha*-*-freebsd* configuration patch

[Loren James Rittle]

Loren> By default, I think this alpha port throws an exception on overflow
Loren> whereas i386-*-freebsd* does not.

> Sounds that way.

OK, see last e-mail for a detailed analysis of the failure.

Loren> Today it is now over 100.
Loren> [http://gcc.gnu.org/ml/gcc-testresults/2002-03/msg00284.html]

> This looks like systematic failure of the execution tests.
> What does libjava.log say?

There is no interesting information there.  Spot checking 10 of the
new failures under gdb, I see this mysterious backtrace:

#0  0x28552841 in _methods76 ()
   from /usr/users/rittle/tmp/gcc-build-latour-mainline-0321/i386-unknown-freebsd4.5//libjava/.libs/libgcj.so.3
#1  0x28084bdd in _Unwind_RaiseException (exc=0x80f7f10)
    at /usr/users/rittle/outside-cvs-src/gcc-mainline/gcc/unwind.inc:109
#2  0x28306bcf in _Jv_Throw (value=0x80f7f10)
    at /usr/users/rittle/outside-cvs-src/gcc-mainline/libjava/exception.cc:99
#3  0x28315b47 in _Jv_CheckArrayStore ()
[...]

Ahh, this is in a tree with shared libgcc finally enabled (in
gcc/config.gcc) for this port.  I did nothing about this over in
libjava yet.  Could this be somehow related?

Regards,
Loren