Re: problem with sablevm-1.1.9 xscale

On Wed, 2005-16-02 at 11:14 +0000, Asutosh Gopinath wrote:
> > So there really is a problem.
> > 
> > The best next step you can take is to recompile with
> > --enable-debugging-features.  This will enable all assert()ions and
> > chances are that one of them will fail before the segfault thus possibly
> > giving out some better clue as to what the actual problem might be.
> > 
> > And even if not, then you still want to debug a version of SableVM
> > compiled with -O0 (no optimizations) which is the case only when using
> > --enable-debugging-features.
> > 
[ ... snip a very good bugreport ... ]

> Please help me find a way to run both rxtx and graphics using one build of 
> sablevm :)

Something fishy is going on and I don't (anymore) believe that ie. "real
life brokenness" is broken.  Reading .jar files that are on CLASSPATH
must not depend on sablevm being compiled in debug mode or not!


Re: A1. "does not seem to find classes in jar file"

When you unpack classes to /usr/local/share/sablevm-classpath they
automagically become part of boot class path.  Alternatively you
can use '-p sablevm.boot.class.path.append=/path/to/file.jar' w/o
unpacking.  (we have yet to add support for 'extensions' dir where
you could just drop your .jars).

The CLASSPATH (not boot...) is handled almost purely in Java.  It would
suggest some weird problem with interpreting java code.


Re: A2. "assertion fails"

Can you run sablevm from within gdb and give us stack trace when the
assertion failure happens?

Also Nizar was telling me last week about 'thread_status' not being
properly initialized at thread creation time, but we had no time to
discuss it, so ATM I am not sure whether it's related or not.


Re: B and C

If things fail with assertions in place, then you can expect "anything"
when assertions are removed.  Getting "debug" version working is the
most important and usually the only step to get a fully functional
SableVM.


Cheers,

				Grzegorz B. Prokopski

PS: Altough we don't own any ARM devices it might be possible to debug
the issue further using freely accessible "Development Cluster" of ARM
machines:

	http://handhelds.org/projects/skiffcluster.html

So guys don't be shy and try to help (and have fun doing so :-)
-- 
Grzegorz B. Prokopski           <[email protected]>
SableVM - Free, LGPL'ed Java VM  http://sablevm.org
Why SableVM ?!?                  http://sablevm.org/wiki/Features
Debian GNU/Linux - the Free OS   http://www.debian.org