2 things about 2.3.5b

John Phillips lfs at linux.demon.co.uk
Sun Jun 25 00:57:05 PDT 2000

Simon Perreault wrote

> 1) Does everyone agree that bin86 does not work? Lilo really doesn't compile
> with it, and I think I'm not alone to have this problem. Once, I played with
> the as86/ld86 switches in the Makefile without ever knowing or noting what I
> was doing, and I was getting more or less errors, and then it compiled fine.
> So I thing either
>     a) the Makefile must be edited so that it works properly (I'm sorry not
> to be able to help on this point);
>     b) bin86 must be replaced by linux86. linux86 works fine, and I'd like
> to know what made you (Gerard) change from linux86 to bin86. What are
> bin86's advantages over linux86?

I agree with 1b (but I didn't try to adjust the Makefile, so there may
be the alternative in 1a) as you say.

> 2) Before making procinfo, you need to, in the Makefile, change -ltermcap
> to -lcurses.

I agree with 2 (-lncurses as you later corrected it).

Here are my complete notes on 2.3.5b on these and other matters.

Things that may be "bugs" in 2.3.5b

1.  In the chroot'd section where you log out and mv bash, you need to
mv bashbug as well (or else you leave two bashbug executables, one from
the static compile and one from the chroot'd compile).

2.  In the chroot'd fileutils, mv is mv'd then the cp and rm of mv fails!
I deleted mv from the second mv list and added a "hash -r" to be safe
against bash's hash list remembering where mv was.

3.  The lilo compile doesn't seem to work with bin86 - I reinstated
linux86 (without trying to fix bin86).

4.  The procinfo Makefile needs -ltermcap to be replaced by -lncurses.

5.  When it comes to creating /etc/sysconfig/network, the directory
/etc/sysconfig/ has to be created first.

6.  In the loadkeys script, the ">/dev/null" needs to be "2>/dev/null"
if you want to suppress the extra message.

7.  Looking at the warnings from texinfo, it advises you to complete the 
installation with "make TEXMF=/usr/share/texmf install-tex".  I added 

8.  The warnings from the chroot'd compile of binutils showed that it
was trying to use ldconfig so I moved ld.so to just before binutils
(this doesn't seem to be a problem - I was just minimising the number of
"errors and warnings").

9.  I created the utmp file just before compiling shadow since this
warned of not finding utmp (this doesn't seem to be a problem - I was
just minimising the number of "errors and warnings").

10. The first boot of LFS 2.3.5b generates an error because there's no
/etc/mtab.  I put a "touch  /etc/mtab" in my script (this doesn't seem to
be a problem - I was just minimising the number of "errors and warnings").

Things that are probably due to my Debian 2.1r5 host

11. Debian's texinfo is old and there's a non-fatal error in compiling
a static bash.  I commented out the call to install-info in doc/Makefile.

12. I still cannot explain why compiling m4 is a problem for me where
it currently is placed.  I have to compile it static and then move the
chroot'd re-compile to after autoconf and automake.

13. I never needed bash-2.03 since I used the patches (although the
patch for glibc is not needed since glibc is compiled with Debian's
rather older bash).

John Phillips           john at linux.demon.co.uk
Mail archive: http://www.pcrdallas.com/mail-archives/lfs-discuss
IRC access: server: irc.linuxfromscratch.org port: 6667 channel: #LFS
Unsubscribe: email lfs-discuss-request at linuxfromscratch.org and put
"unsubscribe" (without the quotation marks) in the body of the message
(no subject is required)

More information about the lfs-dev mailing list