kern/51173: i386 install kernel unable to exec init

[gson%gson.org@localhost (Andreas Gustafsson)]

>Number:         51173
>Category:       kern
>Synopsis:       i386 install kernel unable to exec init
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri May 27 17:25:00 +0000 2016
>Originator:     Andreas Gustafsson
>Release:        NetBSD-current, source date >= 2016.05.20.01.36.16
>Organization:

>Environment:
System: NetBSD
Architecture: i386
Machine: i386
>Description:

The i386 port is still broken on the testbed, and still failing at the
install stage.  Now that kern/51148 is fixed, we run into a new problem,
a failure to exec /sbin/init:

  root on md0a dumps on md0b
  root file system type: ffs
  kern.module.path=/stand/i386/7.99.29/modules
  clock: unknown CMOS layout
  warning: no /dev/console
  exec /sbin/init: error 2
  init: trying /sbin/oinit
  exec /sbin/oinit: error 2
  init: trying /sbin/init.bak
  exec /sbin/init.bak: error 2
  init: trying /rescue/init
  exec /rescue/init: error 2
  init path (default /sbin/init):

Like kern/51148, this only seems to happen with floppy installs.

Tracking down the commit where this problem was introduced was
complicated by the fact that it happened during a time when the kernel
did not even load due to kern/51148, but an automated bisection
patching the tree with the fix for kern/51148 at each step
finally pointed at this commit:

  2016.05.20.01.36.16 christos src/share/mk/bsd.own.mk 1.925

>How-To-Repeat:

Examine the testbed logs, for example:

 http://releng.netbsd.org/b5reports/i386/build/2016.05.27.05.50.07/install.log

>Fix: