[omniORB] omniORB and Red Hat 6.1 segfault

[uche.ogbuji@fourthought.com]

Thanks for the quick response, but I'm still not having any luck.  I have=
 =

spent the last 3 hours trying everything.  First of all, I haven't compil=
ed =

_anything_ except for omniorb.  Everything else comes from RPMs, all of w=
hich =

are standard RedHat issue.

I tried blowing away all executables and libs (note to omniorb developers=
: =

there should be a make option to remove all libs as well as targets: make=
 =

distclean, make mrproper, or something like that).  I've re-compiled it t=
wice. =

I've double-checked paths.  I've done everything.  I still get a segfault=
 =

wnever anything tries to connect to an omniorb server.

I've read the FAQs about egcs versions, but my version is supposedly OK (=
see =

below).  I've looked at the mailing list archives and I saw messages abou=
t =

people who upgraded distributions, but that is not the case with me.  Red=
 Hat =

6.1 is the first OS to have been installed on the hard drive.  I've never=
 done =

an upgrade.

Is there anything else I can check?  Any diagnostics?  Anything?  I'm =

completely at a loss.  I build packages _all the time_ on linux boxes, ev=
en =

packages that use multithreading without such problems.  I have a pretty =
good =

idea what I'm doing, and I'm pretty sure the problem is with omniorb.

One more question: I tried building it with i586_linux_2.0_glibc2.1.mk by=
 =

adding this line to the config.mk.  I still got the segfault.  Is there a=
 =

reason that there is i586_linux_2.0_glibc2.1.mk but no "platform =3D =

i586_linux_2.0_glibc2.1" line in config.mk?

Thanks.

--Uche

> I've seen similar reports with Redhat 6.x when users have upgraded from=

> glibc-2.0 based system (i.e. Redhat 5.2).
> =

> The most likely cause of the problem is that you are linking a library
> (omniORB2 or omnithread runtime) which was compiled on a glibc-2.0 syst=
em
> with the object code that you compile on a glibc-2.1 system.
> =

> Please check that you have rebuilt absolutely everything and no runtime=

> libraries compiled for glibc-2.0 somehow got into your loader path.
> =

> My development machine is now Redhat 6.1, I can assure you there is not=
hing
> wrong there. You have done the right thing in changing the compile
> pathnames.
> =

> =

> Sai-Lai
> =

> >>>>> uche ogbuji writes:
> =

> > I can run the pre-compiled omniORB_280 just fine on RedHat with omniO=
RBpy.  =

> > However, I do get warnings about the size of cout and cerr.  When I t=
ry to use =

> > omniNames, it throws a segmentation fault as soon as a client tries t=
o connect =

> > to it.
> =

> > So today, I tried building the whole thing, using CVS to extract omni=
 and =

> > building it all, including omniORBpy.  After trying this, nothing wor=
ks.  Even =

> > my object servers, which used to run fine with the pre-compiled versi=
on, now =

> > throw a segfault whenever any client tries to connect.  omniNames sti=
ll dumps =

> > core as soon as any client tries to connect.
> =

> > Here is some pertinent data:
> =

> > [uche@medwiredport4 test_suite]$ uname -a
> > Linux medwiredport4 2.2.12-20 #1 Mon Sep 27 10:40:35 EDT 1999 i686 un=
known
> =

> > (note: Red Hat 6.1 for i386 with all errata applied)
> =

> > [uche@medwiredport4 test_suite]$ rpm -qa | grep gc
> > egcs-1.1.2-24
> > egcs-c++-1.1.2-24
> =

> > [uche@medwiredport4 test_suite]$ rpm -qa | grep libc
> > glibc-2.1.2-11
> > glibc-devel-2.1.2-11
> > libc-5.3.12-31
> =

> > I uncommented platform =3D i586_linux_2.0_glibc in config.mk when I t=
ried =

> > building it myself.  I also had to change the paths to the compilers =
in =

> > i586_linux_2.0.mk to
> =

> > CPP =3D /usr/bin/cpp
> > CXX =3D g++
> > CC  =3D gcc
> =

> > Any ideas?  Note that my system is glibc 2.1, but though there is a =

> > i586_linux_2.0_glibc2.1.mk, there is no such option in config.mk.  Co=
uld this =

> > be the problem?
> =

> > Please help: I need to run omniNames today, one way or another.
> =

> =

> =

> =

> =

> -- =

> Sai-Lai Lo                                   S.Lo@uk.research.att.com
> AT&T Laboratories Cambridge           WWW:   http://www.uk.research.att=
=2Ecom =

> 24a Trumpington Street                Tel:   +44 1223 343000
> Cambridge CB2 1QA                     Fax:   +44 1223 313542
> ENGLAND
> =