[omniORB] core> find_solib: Can't read pathname for load map: Input/output error

[Sai-Lai Lo]

Hi! A few points:

1. It is not possible to use gdb to debug multithreaded programs on linux.
   It is not capable of doing that. The same apply to the core file. IMHO,
   this is a major weakness in the linux environment. To support this
   capability requires more support from the kernel via the /proc file
   system.

2. I cannot reproduce the dump core symptom you have reported.  I'm using
   Pentium Pro200, RedHat 5.0, linux-2.0.33, glibc-2.0.7-4,
   binutils-2.8.1.0.23 and egcs-980302.  My guess is the problem is in the
   thread package in glibc. It looks like there is a race condition among
   the threads in handling the SIGTERM. May be you could try updating your
   glibc to see if the problem remains. It is also worth try out
   egcs development snapshot egcs-980302 or later. This version has
   thread-safe exception handling but you have to use omniORB_980410
   because a workaround to a bug in the compiler is needed. We are using
   this compiler on x86 Redhat 5 machines and have compiled some big
   applications successfully.

3. omniORB_980410 is the latest snapshot. I plan to do anyone soon. The
   timing depend on when I can finish merging in DynAny and the testsuite.

Mark Shacklette <mshack@interaccess.com> writes:
> 
> I'm running snapshot_980410 on RedHat 5.0 kernel 2.0.31 with binutils
> 2.8.1.0.26 and glibc-2.0.5c, using the egcs 1.0.2 compiler.  The snapshot
> code is unpatched.  When I run a server, such as eg2_impl or eg3_impl,
> and then run the client, the programs "run" just fine, but when I press 
> Ctrl-C to end the server, I almost always generate a core file.  The core
> file will be generated even if I don't run the client, but simply run the
> server, and then press Ctrl-C.  The core file complains of this:
> 
> Program terminated with signal 11, Segmentation fault.
> find_solib: Can't read pathname for load map: Input/output error
> (gdb) where
> #0  0x8924ebc0 in ?? ()
> 
> I thought that maybe it had something to do with the compiler, so I just
> upgraded to egcs 1.0.3, and rebuilt omniORB_2.5.0, but I still see the same
> results.  If anyone has any idea of what is wrong, I'd appreciate knowing.  I
> like clean executions, and core files make me nervous.
> 
> Thanks for any insights.
> 
> Mark
> 
> PS:  Where can I find the latest omniORB snapshots???
> 
> 
> 

-- 
Dr. Sai-Lai Lo                          |       Research Scientist
                                        |
E-mail:         S.Lo@orl.co.uk          |       Olivetti & Oracle Research Lab
                                        |       24a Trumpington Street
Tel:            +44 223 343000          |       Cambridge CB2 1QA
Fax:            +44 223 313542          |       ENGLAND