Ditto this, except I was trying to finish the build of my MARCH sources
which was previously discussed in this newsgroup (I've been terribly busy
with other things, and had a new hard drive installed, and didn't get NetBSD
transferred to it until just last weekend.)  Anyway, back then it was
suggested that I build gcc before libgcc.  So now I finally tried it, and I
got the exact same problem that Scott mentions below.

BTW, I'm still experiencing headaches with filesystem deadlocks, but I'll
reserve further complaints about that until I try building the new kernel.

At 12:36 pm 5/26/96 -0400, you wrote:
>I have made it through the kernel compile on a V1.1 i386 machine with
>the May 25th tarballs (May 18th died on the pss.c use of ad1848 files
>even though no sound drivers were selected in the config. I posted this
>before but the responses I received did not fix my problem. I added
>sound board support to the config file for the sound boards I don't have
>and it works for now).
>
>This new kernel works fine except that my /usr/src files are stored on a
>nfs drive and nfs can't mount them because of a program version number
>problem when using the V1.1B kernel with my V1.1 userland. (ie the mount
>command for the nfs mount of /usr/src fails). Maybe I just need to add
>an option to my nfs mounts to tell it about using older nfs or
>something.
>
>Anyway, I rebooted my V1.1 kernel and continued on with the V1.1B
>userland rebuild and it has now died on:
>
>===> libgcc
>cc -O -I/usr/src/gnu/usr.bin/gcc/libgcc/../common
>-I/usr/src/gnu/usr.bin/gcc/lib
>gcc/../arch  -I/usr/src/gnu/usr.bin/gcc/libgcc/../arch/i386 -Werror  -c
>-DL__gcc
>_bcmp -o __gcc_bcmp.o /usr/src/gnu/usr.bin/gcc/libgcc/libgcc2.c
>/usr/src/gnu/usr.bin/gcc/libgcc/libgcc2.c:76: no data type for mode `XF'
>/usr/src/gnu/usr.bin/gcc/libgcc/libgcc2.c:82: unknown machine mode
>`__word__'
>*** Error code 1
>
>Stop.
>*** Error code 1
>
>
>In order to get this far I had to rebuild "yacc" first as the %expect
>command was not understood by the V1.1 yacc (it is used in the gcc
>compile now) and I had to rebuild "tsort" as well as the "q" option was
>not understood as well (it is also used in the gcc compile).
>
>If I can get this all rebuilt I will post a shell script which will let
>people who have a plain V1.1 system to try out the impending V1.2
>snapshots. I have this script ready except for this problem.
>
>Here is my file so far for rebuilding. Note that the kernel stuff is
>commented out for now as that part is complete.
>
>#
># i386_v1_1_v1_1b.sh - Upgrade a V1.1 i386 machine to V1.1b (May 25th
>tars)
>#
># Unofficial V1.1 to V1.1B (May 25th) upgrade script...
>#
>#
># Thought I would return the favour (favor in the US) for people trying
>to
># upgrade a straight V1.1 (i386) machine to V1.1B (current) machine in
># prep. for the V1.2 release. Someone gave me this file when I was
>trying to
># upgrade from V1.0 to the -current of the time. So please e-mail me any
># oddities you run into and I will keep reposting this until V1.2 is
>out.
>#
># This procedure has been used to upgrade a raw V1.1 i386 machine,
># without any patches, to a V1.1B machine as of the May 25th tar
># files. Earlier, trying with the V1.1B tar files of May 18th,
># and without any sound boards in the config file, it died on
># an error where pss.c (pss.o) was referring to symbols for a
># particular sound chip even though I had no sound boards in my
># config. This was during the link of the kernel. The compile worked
># fine. I added sound boards to the config file and the kernel
># built fine. If someone knows why this is please e-mail me.
>#
># P.S. If your src directory is nfs mounted you may have to boot your
>old
># kernel to continue as mine was not mounting and complaining of a
>program
># version problem. Maybe something to do with V3 NFS ? After confirming
>my
># new V1.1B kernel worked I just rebooted the original V1.1 and
>continued
># to build V1.1B over NFS like that.
>#
># Scott.Burns@Labatt.Com
># Scott.Burns@Netcontech.Com
>#
>#
>#(cd /usr/src/usr.sbin/config ; make && make install && make cleandir)
>
>#(cd /usr/src/gnu/usr.bin/gas ; make && make install && make cleandir)
>
>#(cd /sys/arch/i386/conf ; config NCTI20_V1_2)
>#(cd /sys/arch/i386/compile/NCTI20_V1_2 ; make depend && make)
># copy  new kernel to /, and boot off it
>
>#(cd /usr/src/share/mk ; make install)
>#
># Rebuild yacc first or the yacc from v1.1 dies on Error Code 1 when
># it hits %expect 34 in /usr/src/gnu/gcc/common/c-parse.y. The
># new yacc included with V1.1B handles it so build and install it
># before it gets used in the gcc building.
>#
>#(cd /usr/src/usr.bin/yacc; make && make install && make cleandir)
>#
># Rebuild tsort first of the tsort from v1.1 dies on Error Code 1 when
># it hits a Make in the gcc build that uses a new "q" option. The
># new tsort included with V1.1B handles it so build and install it
># before it gets used in the gcc building.
>#
>#(cd /usr/src/usr.bin/tsort; make && make install && make cleandir)
>#
># gcc directory was gcc2 ununder v1.1, now it's just gcc.
>#
>(cd /usr/src/gnu/usr.bin/gcc ; make && make install && make cleandir)
>
>(cd /usr/src/bin/sh ; make && make install &&  make cleandir)
>
>(cd /usr/src/include ; make && make install)
>(cd /usr/src/gnu/usr.bin/ld ; make && make install && make cleandir)
>(cd /usr/src/lib ; make && make install && make cleandir)
>
>(cd /usr/src/bin ; make && make install && make cleandir)
>(cd /usr/src/sbin ; make && make install && make cleandir)
>#
># This mkdir was for a bug when going from V1.0 to V1.1, it is probably
># fixed with production V1.1
>#
>mkdir /usr/share/doc/usd/13.viref
>
>(cd /usr/src/usr.bin ; make && make install && make cleandir)
>(cd /usr/src/usr.sbin ; make && make install && make cleandir)
>(cd /usr/src/libexec ; make && make install && make cleandir)
>(cd /usr/src/gnu ; make && make install && make cleandir)
>(cd /usr/src/share ; make && make install && make cleandir)
>#
># If you have the time for games ;-)
>#
># (cd /usr/src/games ; make && make install && make cleandir)
>
>
>Any ideas ?
>
>scott
>scott.burns@labatt.com
>scott.burns@netcontech.com
>
>
>
-----------------------
Trevin Beattie          "Do not meddle in the affairs of wizards,
trevin@xmission.com     for you are crunchy and good with ketchup."
      {:->                                     --unknown