m68k: getting there - slowly

Kiyo Inaba inaba at src.ricoh.co.jp
Mon Mar 29 04:42:15 PST 1999


>It also assumes that the arguments are located at a well-known offset from
>the top of the stack.  This is not true if you save and restore d2-d7/a2-a5
>as I mention above.

Yes, of course.

The other minor problem is current jit code does not construct
proper frame, and when I was in trampoline, gdb can not trace
back.

Rewriting sysdepCallMethod and trampoline codes are on my TODO
list for several months. If you have good alternative codes, I
appreciate it too much.

# Primary target for me is not m68k, by the way :-)

>I find this code a bit too "cute" for my tastes.  May I suggest that when
>someone makes use of the stack layout like this, that at least a comment is
>put in the code, so that I don't break anything later?

I agree. And this is the reason why I put modification history
like
	* Based on the ports
	*      by Remi Perrot <r_perrot at mail.club-internet.fr> to m68k/linux
	* and
	*      by Alexandre Oliva <oliva at dcc.unicamp.br> to sparc

in this file. I also spent several minutes why oliva's implementation
works fine for sparc, at the first time.

Anyway, better comment should be needed here. Is a link to the
'sysdepCallMethod document' (you can get it from Kaffe home
page as you know) is enough?

Kiyo


More information about the kaffe mailing list