🚀 go-pugleaf

     I've issued a couple of queries in recent weeks having to do
with problems I was having bringing up 2.8bsd on our system.
We have a PDP-11/44 with .5 Meg. memory and RM02 drives. It
is finally up and running, and I'd like to thank those who sent me
helpful mail.  The meager documentation included with the tape suggests
that you will need about 8 hours to set things up.  That might have
been true if all the software had been in order.  It wasn't, though.
The tape seemed to be put together in a big hurry.  Makefiles didn't
run, there were typo errors in programs.  This mightn't have been so
tough except that you usually have to come in at odd hours so you can
bring the machine down.  Then you flounder around trying various schemes
in order to get some clues.  I learned a great deal (I've worked with
UNIX less than a year) but it was all the hard way.
     My problem coming up multi-user disappeared when we recompiled
most of the programs in /etc.  In order to do this, however, we had to
fix /lib/libc.a, the C standard function library.  Programs which called
read, write, lseek, etc. wouldn't compile.  Error messages were like;
Undefined;
_lseek
_read
_write
_ioctl

         At first we thought it was the archiver, but then discovered
that all the traps in the '.s' files of the form   'sys   ioctl'  were
left unresolved.  These were all defined in /usr/include/sys.s but were
compiled as local absolute instead of global absolute.  I don't know
why.  We wound up manually inserting the 'ioctl = 43.' stuff into the
proper places.
     Now another problem.  We'll be running along okay, then do a
routine 'fsck /dev/disc0' to make sure the root filesystem is in good
shape.  Fsck starts checking blocks and sizes and then prints;
   PARTIALLY ALLOCATED INODE=1
	   CLEAR?(Y/N)

     Well, if you clear your first inode then you are in trouble.
I haven't chased this one down yet and it doesn't speak well for the
integrity of the system as a whole.  Any suggestions?