[kaffe] kaffe awt and thread management on MIPS

Gianluca Moro giangiammy at yahoo.com
Fri Sep 30 08:15:48 PDT 2005


Hi,

I compiled my kaffe on mips with the following 

./configure --with-jikes=no --with-engine=intrp --enable-pure-java-math
--without-esd --without-kaffe-x-awt --with-kaffe-nanox-awt
--with-rt-jar=/home/AMDgiammy/kaffe-1.1.5/libraries/javalib/rt.jar
--without-classpath-gtk-awt --enable-debug
--with-nanox-include=/usr/include/microwin
--with-nanox-library=/usr/lib --with-threads=unix-jthreads

perhaps the problem on threads in the MIPS porting
can be solved with ad appropriate configuration of
./configure or CFLAGS or so: I found in the web, for
example

setenv CFLAGS "-G 0 -mno-abicalls -fno-pic -mcpu=r4300 -mips2" 

but such a configuration fails in the configuration, while finding
the size of a void *.

Is there someone who compiled a more recent version of kaffe
(I'm working on 1.1.5), and have a working configuration?
(I doubt to be able to find a bug on the config/mips/* side :-)

thanks
giammy 

--- Dalibor Topic <robilad at kaffe.org> wrote:

> Gianluca Moro wrote:
> > I'm testing kaffe in MIPS platform with Nano-X: my test program
> > open a windows, but do not show anything in it (while the same
> > program on Pc show the inner objects)
> > 
> > Tracing the code, I see that the function
> > kaffe-1.1.5/libraries/javalib/java/lang/VMThread.java:sleep() 
> > is calling the function vt.wait with parameter 250ms,
> > but it never returns.
> > 
> >>From what I understand, that function calls 
> >
> /kaffe-1.1.5/libraries/clib/native/Object.c:java_lang_VMObject_wait,
> > whose debugging message gives me
> > 0x10276010 (0x102622f8) waiting for 0x102622f8, 0
> > 
> > It seems that here the parameter (250ms) became 0:
> > do I understand correctly?
> > And if so, why (and where) the 250 value was lost?
> > (kaffe is ocmpiled with unix-jthreads)
> 
> Interesting. That might suggest there is a bug in the config/mips/*
> implementation of the ABI, or in the JNI code.
> 
> You can try bulding Kaffe with libffi, which comes from gcc, and is
> in
> general up to date with different ABI changes. If it's a bug in the
> JNI
> code, you could look at fixing the compiler warnings in
> kaffe/kaffevm/jni* directory.
> 
> cheers,
> dalibor topic
> 


--
Gianluca Moro             Visit  http://ilpinguino.altervista.org/
giangiammy at yahoo.com      MyBlog http://blog.libero.it/giangiammy/


		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com



More information about the kaffe mailing list