[kaffe] Status of m68k/{linux, netbsd}

Kiyo Inaba inaba at src.ricoh.co.jp
Sat Apr 23 04:40:54 PDT 2005

For the last several months, I got several hardware failures for my
test benches. First my Mac-IIci was broken and tried to find new replacement.
Fortunately, I got IIci in Akihabara, replaced harddisk (which was
configured for netbsd-1.6) and continue testing kaffe. But then, I
got another failure of IIvx. I love IIvx except for the case, by the
way ;-) I made new power cable for harddisk, and boot it again. Testing
for linux can be started...

And of course between doing so, I made two business trips, one to
US, and the other to Lisbon.

Well, ok, these are not the issues others are so interested. Going back
to the thing what should be discussed here. Currently, I got both linux
and netbsd working, and I started to test why I have to set optimization
of gcc to be '-O0' rather than '-O2'. It is only set for netbsd and
last time I tried, I got errors for this configuration.

I first check by cross compiling, and even '-O2' works ok. So I submit
usual weekly test, and finally I noticed '-O2' does not work for
m68k/netbsd/jit. The cross compiling test was done for jit3 rather
than jit.

So, I plan to put the compilation flag back to '-O2' (eventually, just
remove '-O0' from config.frag), but it may make jit useless. Is it better
to keep some documentation for this in somewhere?

One additional note, for m68k/netbsd '-O1' is faster than '-O2' (or
very close), and as I reported before, jit is faster than jit3. This
is another issue what should be mentioned.

Any comments?

P.S. When testing m68k becomes more stabilized, I will restart sh3/jit3
     project :-)

More information about the kaffe mailing list