Messages in this thread | | | Date | Mon, 15 Feb 1999 17:53:12 -0500 | From | Arvind Sankar <> | Subject | Re: Possible ld-so bug. |
| |
On Sun, Feb 14, 1999 at 10:23:51PM -0500, Richard B. Johnson wrote: > > > The fact that `make` and `make test` do not produce any errors > means nothing! `make install` can (read will) destroy your > system. I don't know if the bug is in ld.so or in glib-c, but > with a new 'C' compiler, the generated code will have different > offsets which means that new and old libraries can't be mixed. > Somebody (not me) mixes them, the result being a disaster.
I don't get it. Why would different offsets in the libraries make any difference to programs (or other libraries) ? Aren't they supposed to refer to nice symbol tables in libc etc?
FWIW, I installed glibc-2.0.112 over glibc-2.0.111 over glibc 2.0.7pre6 without trouble. The only weird thing was that immediately after installing, you cannot execute any dynamically linked executables (which at least on a redhat system essentially means everything but ldconfig: su, login, even init) because the make install of glibc doesn't run ldconfig. This was a problem with the glibc-2.0.7pre6 to glibc 2.0.111 install (I think it changed ld.so to point somewhere else, but not the C libraries etc). But you run ldconfig, and everything's fine.
-- arvind
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |