I've been trying to work around the problems where the system can't find
init and breaks to the debugger. 

This is a performa 550 without FPU running Kanjitalk (Japanese System)
7.5.5, I'm trying to run nbsd 3.0.1. Attempting to sysinst gives the
usual stuff about failing to mount the system, so I tried the
traditional method instead of the workaround. 

I did not think to disable the system extensions during the install, but
I have disabled them at this point. (Manually moved everything out of
the extensions folder and control panel folder. Extensions manager moved
a bunch of stuff out for me, but left a bunch of things that I imagine
are not supposed to load by themselves, except the extensions for
displaying kanji did load until I removed them.)

Even with the traditional method, I had to use the mini-shell, mount the
partitions by hand, and use the fstab force command to write /etc/fstab.

I can now boot the installer kernel and mount everything under
targetroot, run fsck -f to make sure the file system is (theoretically)
intact. the init not found occurs with both straight and sbc kernels,
booting single user or multi.

    ls -l /targetroot/sbin/i* 

shows an init to be present there, size is 22002, owned by wheel:root,
permissions -r-xr-xr-x .

    ls -l /targetroot/libexec

shows (Copying by hand because I don't have time today to rig a
terminal. Does copy/paste in the mini-shell work even though it's grayed
out in the menu?)

-r-xr-xr-x   1 root  wheel  51265 Jul 14 00:16 ld.elf.so

And

    ls /targetroot/lib

shows

libc.so 
libc.so.12 
libc.so.12.128.2
libcrypt.so
libcrypt.so.0
libcrypt.so.0.2
libcrypto.so
libcrypto.so.2
libcrypto.so.2.1
libedit.so
libedit.so.2
libedit.so.2.9
libevent.so
libevent.so.0
libevent.so.0.2
libipsec.so
libipsec.so.2
libipsec.so.2.1
libkvm.so
libkvm.so.5
libkvm.so.5.2
libm.so
libm.so.0
libm.so.0.2
libradius.so
libradius.so.0
libradius.so.0.0
libtermcap.so
libtermcap.so.0
libtermcap.so.0.5
libutil.so
libutil.so.7
libutil.so.7.6
libz.so
libz.so.0
libz.so.0.4

reboot dumps me to the debugger on a uvm_fault type 8, MMU fault. (If
the numbers are important, I could try to capture this to a terminal
next weekend.)

The actual errors when I try to boot from the kernel on root, after it
says the file system type is ffs (again, copied by hand), 

exec /sbin/init: error 8
init: trying /sbin/oinit 
exec /sbin/oinit: error 2
init: trying /sbin/init.bak
exec /sbin/init.bak: error 2
init: not found
panic: no init
Stopped in pid 1.1 (init) at    netbsd:cpu_Debugger+0x6:      unlk    a6
db>

and attempting to reboot from the debugger will panic again.

I'm out of time to try re-installs this weekend, but if someone knows a
magic word or two to jump me from here past this, I'd appreciate it.