[lfs-support] Errors compiling glibc in chapter 6
zarniwhoop at ntlworld.com
Tue Jun 5 15:25:22 PDT 2012
On Tue, Jun 05, 2012 at 10:49:01PM +0200, joao.moreira at free.fr wrote:
> Hi all,
> This is my second attempt at LFS, many years ago I successfully got counted as number 7178 with LFS 4.0 :-)
> Now I've followed the stable 7.1 version of book, even though I knew from the outset that my host did not conform to the requirements, as you can see in the attached file. The host's gcc and glibc are too recent, and have not been tested... well, maybe now I *have* tested them, and they don't work, I don't know. Oh well, I tried. This is a Fedora 16 system on an HP Pavilion dv7 (Core i5, 64 bits). I followed the book without any issue until installing man pages in chapter 6. Then I get a problem with glibc.
I *doubt* that newer versions of gcc and glibc are the problem.
But, I do remember that sometime in the past newer versions from
redhat or fedora did cause problems.
> 2. Then I start building it and get many many errors such as :
> make: *** No rule to make target `/sources/glibc-build/iconv/stamp.o', needed by `/sources/glibc-build/libc.a'.
That's probably the key, but I don't recognise the error.
> 3. This is followed by many more messages such as :
> make: Target `tests' not remade because of errors.
> 4. It finally ends with
> make: Target `check' not remade because of errors.
> make: Leaving directory `/sources/glibc-2.14.1'
> make: *** [check] Error 2
3 and 4 seem normal if the tests failed. ISTR we didn't run 'make
check' before LFS-5.0 - quite a lot has changed since you were last
> Actually, I don't even know if these are fatal errors, maybe I should try to continue from here...
I would expect 'make check' to run tests. Many of them. In my
opinion, a few errors when *running* the tests are probably ok,
but dozens of them are not.
> Any help or suggestions are welcome (but I don't think I can downgrade the host's glibc just to conform the book's requirements),
> Details of the errors :
That doesn't seem to be the output from
'grep Error glibc-check-log' -
if you have that file, please run that command.
Did you use a fresh glibc-build directory, and freshly untarred
source ? (same question for each build of binutils and gcc).
FWIW, my own logs from 7.1 do NOT mention iconv/stamp.o
Did you run 'make' before you ran 'make check' ?
For the moment I think something is wrong in your build, but I've
no idea what is different.
> [root at hotel lfs]# ./version-check.sh
> bash, version 4.2.10(1)-release
> /bin/sh -> /bin/bash
> Binutils: version 22.214.171.124.1-6.fc16 20110716
That might give problems compiling 3.4 or later kernels in fedora,
(or not - I know *some* hjl-binutils versions have given problems
after some checks in the kernel buildscripts were tightened up) but
is unlikely to screw up LFS.
> bison (GNU Bison) 2.5
> yacc not found
> bzip2, Version 1.0.6, 6-Sept-2010.
> Coreutils: 8.12
> diff (GNU diffutils) 3.2
> find (GNU findutils) 4.5.10
> GNU Awk 4.0.0
> /usr/bin/awk -> /bin/gawk
> gcc (GCC) 4.6.3 20120306 (Red Hat 4.6.3-2)
> (GNU libc) 2.14.90
> grep (GNU grep) 2.9
> gzip 1.4
> Linux version 3.1.0-7.fc16.x86_64 (mockbuild at x86-07.phx2.fedoraproject.org) (gcc version 4.6.2 20111027 (Red Hat 4.6.2-1) (GCC) ) #1 SMP Tue Nov 1 21:10:48 UTC 2011
> m4 (GNU M4) 1.4.16
> GNU Make 3.82
> patch 2.6.1
> Perl version='5.14.2';
> GNU sed version 4.2.1
> tar (GNU tar) 1.26
> Texinfo: makeinfo (GNU texinfo) 4.13
> xz (XZ Utils) 5.1.1alpha
> gcc compilation OK
> [root at hotel lfs]#
Those seem reasonable, although as you say gcc and glibc are newer
than what we've tested. I think you are doing something different
in how you build things (questions about glibc-build, and make vs
make check are above). If there is a problem with those versions of
gcc, glibc, and indeed binutils, I would expect to see more test
failures than the rest of us. For the moment, I don't think you've
been able to *run* the tests.
das eine Mal als Tragödie, das andere Mal als Farce
More information about the lfs-support