>Also, programs compiled from fresh sources (and header files) likely won't
>run on older kernels.  Try to build a new kernel first from an up-to-date
>/usr/src/sys directory.  The existing `config.new' should work fine.  Perhaps
>building genassym is a bit of a problem, if so, try building with assym.s
>included below which was generated on my machine only a few days ago.  Use
>at your own risk, though ...

For what it's worth, starting with Theo's April 25th binaries and completely
-current source (I sup nightly), I was able to build a kernel and a libc.a
after having only done a "make install" in /usr/src/include first.  No
problems at all as far as I could see.  No complaints from building genassym.

Unfortunately, that's as far as I got before leaving work, and now that I
see from last night's sup scan that half the universe has changed :-) I
suspect it might be for naught.  Gotta learn to start upgrade attempts when
I *first* get into work :-)

Any comments on the ramifications of last night's changes?  If I continue on
with building a new ld/gcc/gas and reboot with my yesterday's kernel, and
then re-build userland, will I run into anything else that depends on all
those filesystem kernel changes being installed to work?  Is there any
likelihood that the 4.4BSD-lite filesystem changes will cause any future
problems for folks who've used the SunOS "newfs" to create their initial
UFS filesystems?

	- Greg


------------------------------------------------------------------------------