[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: installing glibc 2.1.x and 2.2.x in parallel?



Warning; the below post might be better on -user, or -powerpc, except that I
have cleverly managed to erect a thin veil of relevance that ties it in to the
question I am replying to. I posted a post about this saga on -powerpc
yesterday, but have gotten no responses, whereas I have tried very hard today
to get the installation working, yielding the problems described below.

On Fri, Jul 06, 2001 at 10:39:09AM -0700, Daniel Jacobowitz wrote:
> Use LD_LIBRARY_PATH, and instead of running ./executable, run
> /foo/lib/ld-2.1.3.so ./executable.  I hope you don't want that program
> to exec() anything else, though.

What do you do if the program may exec something else? I may be having
problems related to this in attempts to install Debian on a rather resistant
G4. For example, assume I'm running off of a CD-ROM burned from a CVS build
of boot-floppies from 30 Jun 2001. This uses busybox and an old libc. I want
to chroot into my mounted filesystem and open up a shell and have the ability
to install packages, compile kernels, etc. -- in other words, do everything --
without some packages giving me segfaults, illegal instruction errors (sig 4),
shared library errors, and obscure errors I've never seen before having to do
with something called Verneed sections. (All of this actually happened. It's
very interesting and frustrating at the same time. The misbehaving programs are
tar, cc1 (in gcc 2.95), and perl 5.6.1; I doubt the problem is in the program;
the way I fixed tar was to download the .deb with busybox wget from outside
the chroot, extract the tar program with busybox ar/tar/gzip, and copy it over
the installed tar program. I didn't have the guts to mess with perl.) How can
I do this? This obviously execs other programs, so the solution you described
presumably doesn't work.

By the way, the reason I don't boot directly into the installed system is
because after freeing unused kernel memory and recognizing my usb hardware,
the kernel all but hangs. (i.e., the boot stays stopped but if I unplug and
reinsert my USB hardware there are corresponding messages showing up from the
kernel, so I know it is not hung.) I tried to compile a new kernel, but had
many problems with that; even though I installed ncurses, cc1 got signal 4 as
mentioned above, so the menu interface could not be used for configuration. For
some reason, the help wasn't available when I finally resorted to the venerable
line-based program.

I am almost ready to forget the whole thing and rest content with Mac OS X on
this machine (which is my main work machine at the office). But I'd prefer
Debian, obviously.

- Jimmy Kaplowitz
jimmy@debian.org

Attachment: pgpmW7j38ry0n.pgp
Description: PGP signature


Reply to: