[omniORB] Anybody seen a crash cleaning up thread specific storage like this?

[Jonathan Biggar]

Here's the gdb output from an application running OO 4.0.5 on Linux with
a 2.6 kernel:

#0  0x002557a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
#1  0x002967d5 in raise () from /lib/tls/libc.so.6
#2  0x00298149 in abort () from /lib/tls/libc.so.6
#3  0x002ca27a in __libc_message () from /lib/tls/libc.so.6
#4  0x002d0abf in _int_free () from /lib/tls/libc.so.6
#5  0x002d0e3a in free () from /lib/tls/libc.so.6
#6  0x00263a5d in ___tls_get_addr_internal () from /lib/ld-linux.so.2
#7  0x0036bf7b in __libc_dl_error_tsd () from /lib/tls/libc.so.6
#8  0x00261045 in _dl_catch_error () from /lib/ld-linux.so.2
#9  0x0036bbce in __libc_dlsym () from /lib/tls/libc.so.6
#10 0x004242f0 in _Unwind_ForcedUnwind () from /lib/tls/libpthread.so.0
#11 0x00421f81 in __pthread_unwind () from /lib/tls/libpthread.so.0
#12 0x0041df00 in pthread_exit () from /lib/tls/libpthread.so.0
#13 0x082ef62b in omni_thread::exit ()
#14 0x082eed86 in omni_thread_wrapper ()
#15 0x0041d341 in start_thread () from /lib/tls/libpthread.so.0
#16 0x00335fee in clone () from /lib/tls/libc.so.6

The application crashed with this message:

  *** glibc detected *** free(): invalid pointer: 0xb7f3db88

I saw this crash once or twice when running my application under VMWare,
but I finally saw it yesterday without virtualization.  I think it's
happening when the application is trying to exit, but it's hard to
capture much detail at that point.

-- 
Jon Biggar
Levanta
jon at levanta.com