[kaffe] MIPS Support Status - New Regression Failures

[Kevin D. Kissell]

I've verified that the" bad FP value" problem in building classfiles
still exists if one tries to make kaffe from current CVS sources for 
MIPS with JIT3 enabled.  More disturbingly, some of the regression
tests are now failing in the interpreter build, specifically GCTest.java 
and ProcessClassTest.java Both of those were passing with an intrp
build from the October 3 2002 CVS tree.  Are they running now 
for all other architectures but MIPS???

I note that the main fix I'd made to the MIPS JIT support,
declaring all argument passing registers to be Reserved in config/mips/jit.h
(including the floating-point registers) seems to already have been
checked in - that's great.  That allowed the 1.0.7+ build to pass
the NativeMethod test in a jit3 build.  I'd made one other functional fix, 
but that was only to correct some misleading jit3 VMDEBUG output,
and I'd added some comments and cleaned up some formatting,
but I'll need to re-implement those patches against the CVS sources,
as they affect files that were also touched in the PS2 update of a few
months ago. 

            Regards,

            Kevin K.