Re: libjava testsuite hanging on OpenBSD

[Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>]

In article <20020529135637.A25482@tetto.liafa.jussieu.fr> you write:
> I've got various parts of the libjava testsuite hanging at runtime
> (Array_3, for starters), and it seems that the dejagnu framework
> doesn't catch it (timeout = 300 ? what does that mean ? it should abort
> after 5 minutes, right ? Array_3 went to 24 minutes before I killed it).

Yes, it should have killed the test after 5 minutes.

> The libjava suite [results are] abysmal so far... [...]

> or maybe I haven't the right dejagnu version ? is 1.4.2 enough ?

I use 1.4.[01] at the moment, so that should be OK.

Having recently gone through all this for FreeBSD on i386 and alpha, I
can offer some advice.  Before investigating any particular issues in
libjava, look closely at libffi (not tested by default) and boehm-gc
(not reported in summary) test results.  Here is how to do it:

$ cd <target>/libffi
$ ./ffitest
[...]
Looks good. No surprises.
$ cd <target>/boehm-gc
$ gmake check
Completed 3 tests
[...]
PASS: gctest
==================
All 1 tests passed
==================

If your port doesn't produce results similar to the above, then fix
that first (boehm-gc affects all executable tests; libffi less so).

Assuming that you have similar results for the support libraries,
given the failure count you posted, I suspect you have at least one
systematic port issue causing most/all failures.  You will need to
look at the output in <target>/libjava/testsuite/libjava.log and
probably run failing tests under the debugger.

If they make no sense to you, report any exact systematic symptoms to
the java list.

Regards,
Loren
-- 
Loren J. Rittle
Senior Staff Software Engineer, Distributed Object Technology Lab
Networks and Infrastructure Research Lab (IL02/2240), Motorola Labs
rittle@rsch.comm.mot.com, KeyID: 2048/ADCE34A5, FDC0292446937F2A240BC07D42763672