[lfs-support] LFS 7.10 - libreadline.so.6 not found error - when running make in 6.40 Perl-5.24.0
lampovas at gmail.com
Mon Apr 9 01:45:40 PDT 2018
On 8 April 2018 at 21:34, Paul Rogers <paulgrogers at fastmail.fm> wrote:
> @Nikos, I am currently running 7.10 (albeit an x86-64 build), and still
> build/maintain my 32-bit (B)LFS systems. I hadn't been paying attention to
> your thread, currently distracted setting up the 400+ scripts of my 8.1
> build. I'm going through your thread "with gun and camera". I don't have
> any of the logs from that build, they're only for debugging a package's
> build. Let's see what shows up.
> Thanks a lot :)
> But first, IF it's necessary to start over again, I think it may be worth
> the effort to upgrade gcc in Ch-5.5, -5.10, & -6.17 to 7.3 to get retpoline
> and hope there's no breakage. Then I'd also patch the kernel up to get the
> KPTI patches. I've patched up to 4.9.75 to get some early patches, and
> plan to keep up with the latest 4.9 patches for now. I followed Ken's
> approach of a separate gcc-7.3 in /opt for compiling the kernel.
> Now to your thread...
> - I have some reservations about using Virtual Box, and Ubuntu though
> others say that works (I can't. I've been using LFS hosts since 4.1). One
> must be certain of the image VB is presenting your build environment. In
> particular for a 32-bit build, there's the problem of building gmp for
> 32-bits on a 64-bit system. I find I must be "forensic" about
> cross-building like that. I use --target quite a lot, just to make sure
> gcc knows what it will run on.
Sorry for not understanding, when u mean 64-bit system u refer to the VM
host which is Ubuntu 12.04 32-bit? Or my laptops OS which is windows 10
64-bit? i don't get it why is it a problem, since my VM runs on 32-bit.
- I agree with Thanos that the "ldd /bin/bash" is a major problem, and
> means you must backtrack at least as far as 6.33, and with your diagnosis
> that there was already a problem at 6.17. "ldd /usr/bin/file" shouldn't
> still be looking at /tools, which strongly implicates 6.10, "Adjusting the
A strange thing that i have not noticed is this.
Binaries 6.12 - 6.16 compiled with the old gcc have the correct paths on
shared libraries on ldd.
Thus because i keep vm snapshots, i skipped building 6.17 gcc and moved to
Then 6.18 Bzip2 without having build the 6.17 gcc, had the correct paths
for shared libraries on ldd.
Thus the problem lies on 6.17 gcc.
Then I reverted the snapshot with the 6.17 gcc been build and i inspected
I found these references with 'tools' that i didnt like that much..
COLLECT_GCC_OPTIONS='-v' '-mtune=generic' '-march=pentiumpro'
--eh-frame-hdr -m elf_i386 -dynamic-linker /tools/lib/ld-linux.so.2
-L/tools/lib /tmp/cc5hD6yA.o --verbose -lgcc --as-needed -lgcc_s
--no-as-needed -lc -lgcc --as-needed -lgcc_s --no-as-needed
And then another strange thing is that when i inspected the dummy.log of
6.10 i found that in
the option -dynamic-linker is pointing to /lib/ld-linux.so.2 and not
-dynamic-linker /tools/lib/ld-linux.so.2 as it is in 6.17 gcc.
> (And in the absence of a good package manager watching your build and able
> to selectively remove everything since, I hope you tarballed Ch5 at
> completion, and can start over with a clean "partition" at the start of
> Ch6. I always do that and keep it until the build is running reasonably
i keep snapshots for every group of packages been built (i split section
in 8 groups of packages), i revert them and i dont reboot.
> - I think where you made your mistake is:
> > there is a reference of tools here but is accepted according
> > to this log
> > http://lfs.phayoune.org/lfs/build-logs/8.1-rc1/pentium4/
> That is NOT authoritative. The book is!
> Your SEARCH_DIR still references /tools. Stop there. Until that is
> fixed, everything else is a waste of time and effort.
> You need to get back to the end of 6.9 with everything "squeaky
> clean"--which may not be easy, but you must be able to prove it--and redo
> 6.10 exactly as the book says. If it doesn't work, then something has
> already departed from the book's path--and that's why I used the word
> "forensic" above. No deviations are allowed. And yes, from time to time
> I've had problems there myself.
> - I start all my build scripts "#!/bin/bash -e" so they "drop out" on
> errors. Of course 6.10 relies on visual inspection. I intersperse "echo
> "Continue?"; read dummy" at each check, and ^C if it doesn't match.
> Paul Rogers
> paulgrogers at fastmail.fm
> Rogers' Second Law: "Everything you do communicates."
> (I do not personally endorse any additions after this line. TANSTAAFL :-)
> 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?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the lfs-support