Rebuilding RHEL3 libc.so from source

[Michael ODonnell]

Jarod Wilson (et al) wrote:
 [much useful stuff]

Thanks for the feedback.  I now have a system sort of running with
libraries installed from a locally built RPM.  It's at least running
well enough that the app developer in question can probably make progress
debugging his stuff, but there's still plenty that I wish I had a better
handle on.  For example: the RPM delivered by RHAT is named

   glibc-2.3.2-95.50.i686.rpm

...but when I rebuild from (what I assume to be) this corresponding SRPM:

   glibc-2.3.2-95.50.src.rpm

...these are all that end up in the output hierarchy:

   src/rpm/RPMS/
   src/rpm/RPMS/athlon/
   src/rpm/RPMS/i386/
   src/rpm/RPMS/i386/glibc-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-common-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-debug-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-debuginfo-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-debuginfo-common-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-devel-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-headers-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-profile-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/glibc-utils-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i386/nscd-2.3.2-95.50.i386.rpm
   src/rpm/RPMS/i486/
   src/rpm/RPMS/i586/
   src/rpm/RPMS/i686/
   src/rpm/RPMS/noarch/

...so I suspect that may have something to do with the differences that
I mentioned between the libs I'm building versus what's in the RHAT-built
RPM.  Can somebody tell me how to persuade rpmbuild to generate anything
named *.i686.rpm from that SRPM?

FYI, the "arch" command on the box where I'm building says "i686" and
uname -a says:

   Linux hooha 2.4.21-47.EL-smp #6 SMP Mon Nov 13 16:14:24 EST 2006 i686 i686 i386 GNU/Linux

...and I haven't identified anything in my spec files that has the
desired effect.  I've been trying to RTFM in places like these:

   http://docs.fedoraproject.org/drafts/rpm-guide-en/
   http://www.ibiblio.org/pub/Linux/docs/HOWTO/other-formats/html_single/RPM-HOWTO.html

...which are pretty good docs, but no joy, so far.

It's those *.i386.rpm packages I built that I've installed on the
target machine - they mostly work but there are some strange problems
with unresolved symbols that I'll probably gripe about in another msg.

(and I'll also describe a cool dynamic-linker debugging trick)