In article <Pine.NEB.4.58.0403070914250.2259@seduction.immanent.net>
fredb@immanent.net wrote:

> I'm not following how you came to the conclusion that gcc-2.95.2 is
> involved. It looks like the host is running NetBSD-1.6ZK-m68k.

It shows only kernel version, and I guess most of (slow) m68k machines
are not updated so frequently...

> There have been a lot of changes to the toolchain recently, so it
> may well be that something is borked. FWIW, releng's last successful
> autobuild was from sources dated March 03, 0:00 UTC:

Maybe the problem is m68k specific, and the releng is i386, isn't it?
My guess is:
- gcc-3.3.2 for m68k compiled by gcc-2.95.x on m68k has problems.
- gcc-3.3.2 for m68k compiled by gcc-3.3.2 on m68k has another problem.
  -> see src/doc/HACKS and
     http://mail-index.netbsd.org/port-hp300/2003/11/29/0001.html
     (This means self-build on m68k has been broken for long time.)
- gcc-3.3.2-pre20040209 for m68k compiled by gcc-3.3.2-pre20040209 on m68k
  seems to work.
  -> I've tested it on my hp300
- gcc-3.3.3 for m68k compiled by gcc-2.95.x on m68k also has problems.
  -> I saw this on my news68k, and maybe the same problem with Shibagaki's
I'm trying native build on my news68k with gcc-3.3.3, but I think it will
take about two weeks...
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp