This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: Mauve harness failure on s390
- From: Mark Wielaard <mark at klomp dot org>
- To: Ulrich Weigand <weigand at immd1 dot informatik dot uni-erlangen dot de>
- Cc: raif at fl dot net dot au, java at gcc dot gnu dot org
- Date: 31 Jan 2003 16:41:51 +0100
- Subject: Re: Mauve harness failure on s390
- Organization:
- References: <200301311508.QAA05916@faui11.informatik.uni-erlangen.de>
Hoi,
On Fri, 2003-01-31 at 16:08, Ulrich Weigand wrote:
> What I'm wondering is why this is apparently platform-dependent;
> I'd have thought the compilation to bytecode would be independent
> of the platform, and likewise the question whether any given
> bytecode sequence contains unreachable code or not ...
It isn't platform dependent. I just updated Mauve from CVS and see the
same thing on i686-pc-linux-gnu.
It depends on how gcj uses the Mauve testsuite. The dejagnu test wrapper
fails when it sees unexpected warnings. The unreachable bytecode
warnings should be harmless (although they do indicate that gcj -C
produces non-optimal bytecode). Somehow the libjava_prune_warnings from
libjava/testsuite/lib/libjava.exp isn't working correctly.
> Any suggestions how to debug such problems? Is there something
> like a bytecode disassembler / verifier / analyser tool?
You can use jcf-dump -c to decompile and gcj --syntax-only to verify a
class file.
Cheers,
Mark