[kaffe] JIT and JIT3 Floating Point failures on m68k platform

[Dalibor Topic]

hi tony,

--- Tony Wyatt <wyattaw@optushome.com.au> wrote:

> Now my question is - where (by which function) is the value "Float.Nan"
> evaluated during the compilation of a .java file by the JIT engine? I have
> now been looking for some days without finding which function is to blame.
> 
> Or is it all performed in kjc? Needless to say, I am using the default
> kjc.jar
> as compiler. But something somewhere is giving different behaviour for
> interpreter and JIT.

.java files are compiled by a Java compiler (i.e. kjc in kaffe's case). The jit
compiler (and interpreter) works on already compiled bytecode files (.class).
It seems that the jit and the interpreter diverge somewhere during the
execution of kjc and let it generate a broken file.

So one (but slow) way of doing it would be to use -verbosecall to compare the
output of kaffe running kjc to compile the test class. I'm not sure if
-verbosecall produces any results with intrp engine, though.

You could also take a look at bytecode/classfile/FloatConstant.java from
http://www.dms.at/kopi/download/kjc-2.1B-src.tar.gz and add some code to
monitor  what gets set or written out.

cheers,
dalibor topic

__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com