[kaffe] Unification of jit-m68k.def and jit3-m68k.def (step 1)
inaba at src.ricoh.co.jp
Thu Sep 2 04:54:27 PDT 2004
Dalibor Topic wrote:
>Kiyo Inaba wrote:
>> As I said, I am now unifing machine specific portion of jit file.
>> Attached is the result for step 1.
>> While doing this merge, I found several errors for both jit and jit3.
>> It can be a good co-product of this effort :-)
>Thank you very much for undertaking the effort ;) So the final result
>will be a unified jit-m68k.def? That's great!
Yes, in the last stage, I hope to make only one 'jit-m68k.def' file
which can be used for both jit and jit3. In that stage, maybe we
will have symbolic link to 'jit3-m68k.def' from 'jit-m68k.def'.
The reason to make them unified is for easy maintenance. As jit
hackers may know, the difference between jit and jit3 is not big.
Mainly, the later treats internal used registers smarter (I don't
say, it makes entire process faster, sigh...)
Later, I will back port floating point emulation code to jit, which
is only in jit3 right now. This effort may benefit non-FPU 68K
CPU's mainly used for embedded application. The other important
target is the half-damaged LC series CPU. (These cpu have broken
emulation trap and for long time, there were no way to use them
with unix-like system. This situation was changed by NB2, because
they provide 'soft-float' port for LC.)
OK, my declaration of dream was finished. Going back to the real
>>> I have to debug -DHAVE_NO_FLOATING_POINT first :-<
# That's what I expect, by the way.
>I've checked your patch in, and I'm looking forward to step 2 ;)
So after this debugging effort, I will go back to do the second step.
>p.s. Ugh, my freshly installed Mozilla Mail had HTML mails on by
>default, and I didn't notice it for a few days. Sigh.
Did you send me mail in this period? I sometimes automatically
delete HTML mail without reading the contents ;-)
More information about the kaffe