[lfs-support] expect5.45.4 compile error in Chapter 5, missing files from glibc suspected, um, no

[Clyde McKenney]

Hi Clyde,

from the information you have posted, you have misinterpreted the
problem. An install-sh file can be referenced in many packages (I
think its use is part of the autotools system (autoconf, automake,
libtool).


configure: creating ./config.status
config.status: creating Makefile
config.status: WARNING: 'Makefile.in' seems to ignore the --datarootdir setting
config.status: executing default commands
chmod: cannot access './install-sh': No such file or directory

That is not fatal.


=== configuring in testsuite (/mnt/lfs/sources/expect5.45.4/testsuite)
configure: running /bin/sh ./configure --disable-option-checking '--prefix=/tools' '--with-tcl=/tools/lib' '--with-tclinclude=/tools/include' --cache-file=/dev/null --srcdir=.
checking for correct TEA configuration... ok (TEA 3.9)


et cetera....then at the end:


gcc \
     -pipe -O2 -fomit-frame-pointer -Wall -fPIC  \
     -Wl,--export-dynamic  \
    -o expect exp_main_exp.o \
    -L/mnt/lfs/sources/expect5.45.4 -lexpect5.45.4 \
    -L/tools/lib -ltcl8.6 \
    -ldl  -lm \
    -Wl,-rpath,/tools/lib \
    -Wl,-rpath,/tools/lib/expect5.45.4
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libutil.so.1, needed by /mnt/lfs/sources/expect5.45.4/libexpect5.45.4.so, not found (try using -rpath or -rpath-link)
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libpthread.so.0, needed by /tools/lib/libtcl8.6.so, not found (try using -rpath or -rpath-link)

These are the errors which *are* fatal. I'll comment on the first
two, but basically after one error for a missing library all bets
are off.

The first missing file is /tools/lib/libutil.so.1 - that should have
been installed by glibc.

The second is /tools/lib/libpthread.so.0 which should also have been
installed by glibc.


/tools/lib/libtcl8.6.so: undefined reference to `***@GLIBC_2.2.5'

OK, I'll comment on this one too, since it is almost certainly
defined in libpthread.so.

Do you have those two files ? The version .so.X* should be symbolic
links to .so files : if you have them, does file report they are
broken (i.e. pointing to something which does not exist) ?

Äžen

Ken,



I have those two files, they exist in /tools/lib, during the expect5.45.4 make, as symlinks to existing lib*-2.27.so files, even though gcc complains that they're not there.  (?)

Could the fact that these errors reference "GLIBC2.2.5", when the book and I are using glibc 2.27, be a hint?
FWIW, my aging memory may be wrong, but I believe I successfully compiled all of Chapter 5 and 6 during an earlier run on this machine (but since gave up the partition during a rebuild done for other reasons), so I'll admit it's not a squeaky clean OOB Mint install.

But then FWIW, I just successfully compiled expect5.45.4 inside a fresh VirtualBox using Manjaro, an Arch distro. So blame Mint...

Oh well, I'll probably reboot to bare metal, or just start living inside VirtualBox, but nailing this issue would be educational.



My version-check.sh reports:
bash, version 4.3.48(1)-release
/bin/sh -> /bin/bash
Binutils: (GNU Binutils for Ubuntu) 2.26.1
bison (GNU Bison) 3.0.4
/usr/bin/yacc -> /usr/bin/bison.yacc
bzip2,  Version 1.0.6, 6-Sept-2010.
Coreutils:  8.25
diff (GNU diffutils) 3.3
find (GNU findutils) 4.7.0-git
GNU Awk 4.1.3, API: 1.1 (GNU MPFR 3.1.4, GNU MP 6.1.0)
/usr/bin/awk -> /usr/bin/gawk
gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609
g++ (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609
(Ubuntu GLIBC 2.23-0ubuntu10) 2.23
grep (GNU grep) 2.25
gzip 1.6
Linux version 4.13.0-37-generic (***@lcy01-amd64-012) (gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.9)) #42~16.04.1-Ubuntu SMP Wed Mar 7 16:03:28 UTC 2018
m4 (GNU M4) 1.4.17
GNU Make 4.1
GNU patch 2.7.5
Perl version='5.22.1';
sed (GNU sed) 4.2.2
tar (GNU tar) 1.28
texi2any (GNU texinfo) 6.1
xz (XZ Utils) 5.1.0alpha
g++ compilation OK



User lfs has the following environment set:

BASH=/bin/bash
BASHOPTS=checkwinsize:cmdhist:complete_fullquote:expand_aliases:extquote:force_fignore:hostcomplete:interactive_comments:progcomp:promptvars:sourcepath
BASH_ALIASES=()
BASH_ARGC=()
BASH_ARGV=()
BASH_CMDS=()
BASH_LINENO=()
BASH_SOURCE=()
BASH_VERSINFO=([0]="4" [1]="3" [2]="48" [3]="1" [4]="release" [5]="x86_64-pc-linux-gnu")
BASH_VERSION='4.3.48(1)-release'
COLUMNS=102
DIRSTACK=()
EUID=1001
GROUPS=()
HISTFILE=/home/lfs/.bash_history
HISTFILESIZE=500
HISTSIZE=500
HOME=/home/lfs
HOSTNAME=Dell-Mint
HOSTTYPE=x86_64
IFS=$' \t\n'
LC_ALL=POSIX
LFS=/mnt/lfs
LFS_TGT=x86_64-lfs-linux-gnu
LINES=60
MACHTYPE=x86_64-pc-linux-gnu
MAILCHECK=60
MAKEFLAGS=-j1
OLDPWD=/mnt/lfs/sources
OPTERR=1
OPTIND=1
OSTYPE=linux-gnu
PATH=/tools/bin:/bin:/usr/bin
PIPESTATUS=([0]="0")
PPID=4290
PS1='${debian_chroot:+($debian_chroot)}\u@\h \w \$ '
PS2='> '
PS4='+ '
PWD=/mnt/lfs/sources/expect5.45.4
SHELL=/bin/bash
SHELLOPTS=braceexpand:emacs:histexpand:history:interactive-comments:monitor
SHLVL=1
TERM=xterm-256color
UID=1001


Sorry for html email, Thanks for advice from all!

-- Clyde

[Ken Moffat]

Post by Clyde McKenney
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libutil.so.1, needed by /mnt/lfs/sources/expect5.45.4/libexpect5.45.4.so, not found (try using -rpath or -rpath-link)
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libpthread.so.0, needed by /tools/lib/libtcl8.6.so, not found (try using -rpath or -rpath-link)
These are the errors which *are* fatal. I'll comment on the first
two, but basically after one error for a missing library all bets
are off.
The first missing file is /tools/lib/libutil.so.1 - that should have
been installed by glibc.
The second is /tools/lib/libpthread.so.0 which should also have been
installed by glibc.
OK, I'll comment on this one too, since it is almost certainly
defined in libpthread.so.
Do you have those two files ? The version .so.X* should be symbolic
links to .so files : if you have them, does file report they are
broken (i.e. pointing to something which does not exist) ?
Ken,
I have those two files, they exist in /tools/lib, during the expect5.45.4 make, as symlinks to existing lib*-2.27.so files, even though gcc complains that they're not there.  (?)

I'm glad they are there, but now I'm puzzled about why gcc is saying
they are not there. That usually means they depend on soemthing
else which is missing (and that wouldbe very odd in chapter 5).

I was going to say that I normally discard /tools after booting, but
on the machine I'm currently using I saved it as /tools.orig.
Running ldd on them only shows (for me)

linux-vdso, libc.so.6, and ld-linux.

and if you don't have those you will get nowhere.

Post by Clyde McKenney
Could the fact that these errors reference "GLIBC2.2.5", when the book and I are using glibc 2.27, be a hint?

Maybe, but I'm out of practice on determining what these versioned
symbols mean. Running strings on my preserved libtcl8.6.so from
/tools I can see ***@GLIBC_2.2.5 although it does
not appear to be in libpthread itself (but the @ might mean the
version is determined by other means).

Post by Clyde McKenney
FWIW, my aging memory may be wrong, but I believe I successfully compiled all of Chapter 5 and 6 during an earlier run on this machine (but since gave up the partition during a rebuild done for other reasons), so I'll admit it's not a squeaky clean OOB Mint install.
But then FWIW, I just successfully compiled expect5.45.4 inside a fresh VirtualBox using Manjaro, an Arch distro. So blame Mint...

If you got through chapter 6 on a previous build then I'm baffled
about what might have changed, unless you ran upgrades on mint.

Post by Clyde McKenney
Oh well, I'll probably reboot to bare metal, or just start living inside VirtualBox, but nailing this issue would be educational.
bash, version 4.3.48(1)-release
/bin/sh -> /bin/bash
Binutils: (GNU Binutils for Ubuntu) 2.26.1
bison (GNU Bison) 3.0.4
/usr/bin/yacc -> /usr/bin/bison.yacc
bzip2,  Version 1.0.6, 6-Sept-2010.
Coreutils:  8.25
diff (GNU diffutils) 3.3
find (GNU findutils) 4.7.0-git
GNU Awk 4.1.3, API: 1.1 (GNU MPFR 3.1.4, GNU MP 6.1.0)
/usr/bin/awk -> /usr/bin/gawk
gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609
g++ (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609
(Ubuntu GLIBC 2.23-0ubuntu10) 2.23
grep (GNU grep) 2.25
gzip 1.6
m4 (GNU M4) 1.4.17
GNU Make 4.1
GNU patch 2.7.5
Perl version='5.22.1';
sed (GNU sed) 4.2.2
tar (GNU tar) 1.28
texi2any (GNU texinfo) 6.1
xz (XZ Utils) 5.1.0alpha
g++ compilation OK

I don't recall seeing the GNU MPFR and GNU MP details for gawk
before, but that all looks ok.

Post by Clyde McKenney
BASH=/bin/bash
BASHOPTS=checkwinsize:cmdhist:complete_fullquote:expand_aliases:extquote:force_fignore:hostcomplete:interactive_comments:progcomp:promptvars:sourcepath

Wow, what a lot (I don't have that, or many of the other variables,
but I don't see anything that looks odd.

Post by Clyde McKenney
BASH_ALIASES=()
BASH_ARGC=()
BASH_ARGV=()
BASH_CMDS=()
BASH_LINENO=()
BASH_SOURCE=()
BASH_VERSINFO=([0]="4" [1]="3" [2]="48" [3]="1" [4]="release" [5]="x86_64-pc-linux-gnu")
BASH_VERSION='4.3.48(1)-release'
COLUMNS=102
DIRSTACK=()
EUID=1001
GROUPS=()
HISTFILE=/home/lfs/.bash_history
HISTFILESIZE=500
HISTSIZE=500
HOME=/home/lfs
HOSTNAME=Dell-Mint
HOSTTYPE=x86_64
IFS=$' \t\n'
LC_ALL=POSIX
LFS=/mnt/lfs
LFS_TGT=x86_64-lfs-linux-gnu
LINES=60
MACHTYPE=x86_64-pc-linux-gnu
MAILCHECK=60
MAKEFLAGS=-j1
OLDPWD=/mnt/lfs/sources
OPTERR=1
OPTIND=1
OSTYPE=linux-gnu
PATH=/tools/bin:/bin:/usr/bin
PIPESTATUS=([0]="0")
PPID=4290
PS2='> '
PS4='+ '
PWD=/mnt/lfs/sources/expect5.45.4
SHELL=/bin/bash
SHELLOPTS=braceexpand:emacs:histexpand:history:interactive-comments:monitor
SHLVL=1
TERM=xterm-256color
UID=1001
Sorry for html email, Thanks for advice from all!
-- Clyde

At this point, I'm none the wiser. Maybe somebody else will
recognize the problem. If not, maybe it will disappear on a fresh
build - but those sort of problems are exceptionally annoying and it
would be nice to understand what went wrong.

ĸen
--
Truth, in front of her huge walk-in wardrobe, selected black leather
boots with stiletto heels for such a barefaced truth.
- Unseen Academicals
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thin

[Pierre Labastie]

Post by Clyde McKenney
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libutil.so.1, needed by /mnt/lfs/sources/expect5.45.4/libexpect5.45.4.so, not found (try using -rpath or -rpath-link)
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libpthread.so.0, needed by /tools/lib/libtcl8.6.so, not found (try using -rpath or -rpath-link)
[...]
I have those two files, they exist in /tools/lib, during the expect5.45.4 make, as symlinks to existing lib*-2.27.so files, even though gcc complains that they're not there.  (?)

As Ken, I'm not sure what to advise. The only diagnostic I can propose at this
point is to add the the "-v" and "-Wl,-verbose" switches to the offending gcc
command: cd to /sources/expect5.45.4/, copy the long gcc command, add the
switches, and pipe the output to a file (then hit enter). Then browse the
output (or post it), looking for paths used by gcc for finding headers and
libraries.

Post by Clyde McKenney
Could the fact that these errors reference "GLIBC2.2.5", when the book and I are using glibc 2.27, be a hint?

I do not think so. "GLIBC2.2.5" is just a tag indicating that the symbol (here
"pthread_specific") is the one defined since the 2.2.5 version of glibc, and
that if a library is found with a previous version of the symbol, it should
not be used...

Post by Clyde McKenney
But then FWIW, I just successfully compiled expect5.45.4 inside a fresh VirtualBox using Manjaro, an Arch distro. So blame Mint...

After yesterday's discussion, I ran a build in a VM with ubuntu, and all went
well. I think if version-check.sh is OK, the build instructions should work...
I may try to install mint in a VM though...


Pierre
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.or

[Pierre Labastie]

Post by Pierre Labastie

Post by Clyde McKenney
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libutil.so.1, needed by /mnt/lfs/sources/expect5.45.4/libexpect5.45.4.so, not found (try using -rpath or -rpath-link)
/mnt/lfs/tools/bin/../lib/gcc/x86_64-pc-linux-gnu/7.3.0/../../../../x86_64-pc-linux-gnu/bin/ld: warning: libpthread.so.0, needed by /tools/lib/libtcl8.6.so, not found (try using -rpath or -rpath-link)
[...]
I have those two files, they exist in /tools/lib, during the expect5.45.4 make, as symlinks to existing lib*-2.27.so files, even though gcc complains that they're not there.  (?)

As Ken, I'm not sure what to advise. The only diagnostic I can propose at this
point is to add the the "-v" and "-Wl,-verbose" switches to the offending gcc
command: cd to /sources/expect5.45.4/, copy the long gcc command, add the
switches, and pipe the output to a file (then hit enter). Then browse the
output (or post it), looking for paths used by gcc for finding headers and
libraries.

Post by Clyde McKenney
Could the fact that these errors reference "GLIBC2.2.5", when the book and I are using glibc 2.27, be a hint?

I do not think so. "GLIBC2.2.5" is just a tag indicating that the symbol (here
"pthread_specific") is the one defined since the 2.2.5 version of glibc, and
that if a library is found with a previous version of the symbol, it should
not be used...

Post by Clyde McKenney
But then FWIW, I just successfully compiled expect5.45.4 inside a fresh VirtualBox using Manjaro, an Arch distro. So blame Mint...

After yesterday's discussion, I ran a build in a VM with ubuntu, and all went
well. I think if version-check.sh is OK, the build instructions should work...
I may try to install mint in a VM though...

Everything seems OK with linuxmint-18.3 cinnamon-64 bit (in a VM)...

Pierre
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Do not top post on this list.

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

http://en.wikipedia.org/wiki

[Paul Rogers]

This post might be inappropriate. Click to display it.