powerpc boot-floppies from current CVS broken

[Matt Porter <mmporter@home.com>]

I've resumed work on the powerpc port.  A releases builds with only minor
tweaks which really amazed me.

Unfortunately, dbootstrap is completely dead.  I tested boot/install
image on headed/non-headed PReP systems.  On both systems dbootstrap/newt
had trouble with the display (note that a bf-2.2.1 image with the same
kernel works fine).

On the headed system, dbootstrap starts after the
installer image is booted and displays the release_notes as expected.
However, when hitting <Enter> to continue at the <Continue> prompt only
sends raw keystrokes to the display.  So, I get "^M^M^M^M^M^Masdf" overlaid
on the newt display when typing <Enter> and random keys.  Input seems to
not be captured by the library.

On the non-headed system, dbootstrap starts after the installer image is 
booted but only clears the screen and can continue no further.

Interestingly enough, on both systems I can pass "init=/bin/sh" to the
kernel, mount /proc manually, then execute /sbin/dbootstrap.  In this
case, both systems are able to run through the installation and I/O
is handled as expected.  There are, of course, lots of little problems
along the way to fix due to the code clean up and new features.

I've looked through the changes to dbootstrap since the 2.2.1 release and
they are pretty substantial making it hard to locate the problem.  Does
anybody with better term-handling knowledge have any ideas what would 
cause dbootstrap to have problems running when being respawned from
inittab versus spawning from ash?  The only thing I can think of is that
ash is initializing the console such that newt works correctly.

Thanks,
--
Matt Porter
mmporter@home.com
This is Linux Country. On a quiet night, you can hear Windows reboot.