Hi,

If I set a callback using set_thread_id, I always get an error while
closing the env. The stack is at the end of the email.
I looked into the bdb source code and it looks like a bug to me. But I
am not sure.

DbEnv::close calls:
cleanup()
env->close(env, flags))

The call to cleanup sets the internal env representation to NULL.
Finally when DbEnv::_thread_id_intercept is called as shown in the
stack below, the env is null and it will always throws an exception.

This doesn't happen if I don't set a callback using set_thread_id. In
that case DbEnv::_thread_id_intercept is not called, instead the
default implementation __os_id is called.


#0 0x415d5ae1 in __kill () at __kill:-1
#1 0x409c876b in raise (sig=6) at signals.c:65
#2 0x415d7062 in abort () at ../sysdeps/generic/abort.c:88
#3 0x4016c28b in __default_terminate () at ./libgcc2.c:-1
#4 0x4016c2ac in __terminate ()
from libdb_cxx-4.4.so
#5 0x4016cbe4 in throw_helper (eh=0x40196e50, pc=0x40117024,
my_udata=0xbfffee38, offset_p=0xbfffee34)
from libdb_cxx-4.4.so
#6 0x4016cd9c in __throw ()
from libdb_cxx-4.4.so
#7 0x40052309 in DbEnv::runtime_error (env=0x0, caller=0x401717e1
"DbEnv::thread_id_callback", error=22, error_policy=1)
at ../dist/../cxx/cxx_env.cpp:589
#8 0x40050591 in DbEnv::_thread_id_intercept (env=0x806e160,
pidp=0xbfffefbc, thridp=0xbfffefc0) at ../dist/../cxx/cxx_env.cpp:226
#9 0x40050176 in _thread_id_intercept_c (env=0x806e160,
pidp=0xbfffefbc, thridp=0xbfffefc0) at ../dist/../cxx/cxx_env.cpp:132
#10 0x40117025 in __env_set_state (dbenv=0x806e160, ipp=0xbffff014,
state=THREAD_ACTIVE) at ../dist/../env/env_failchk.c:240
#11 0x4011a872 in __env_close_pp (dbenv=0x806e160, flags=0) at
.../dist/../env/env_open.c:666
#12 0x400508ce in DbEnv::close (this=0x806e0f8, flags=0) at
.../dist/../cxx/cxx_env.cpp:330
#13 0x40020d2e in NewEnv::close (this=0x806e0f8)
at NewEnv.cpp:50
#14 0x40020be0 in NewEnv::~NewEnv (this=0x806e0f8, __in_chrg=3)
at NewEnv.cpp:35
#15 0x0804e122 in CBdbTest::~CBdbTest (this=0xbffff134, __in_chrg=2)
at BdbTest.cpp:308
#16 0x0804dc50 in main (argc=14, argv=0xbffff234)
at BdbTest.cpp:263
#17 0x415c3627 in __libc_start_main (main=0x804dbc8 <main>, argc=14,
ubp_av=0xbffff234, init=0x804d3d4 <_init>, fini=0x805616c <_fini>,
rtld_fini=0x4000dcc4 <_dl_fini>, stack_end=0xbffff22c) at
.../sysdeps/generic/libc-start.c:129

This is caused by a known bug in the C++ API introduced in DB 4.4. I
will send you a patch directly, since posting diffs to the newsgroup is
not very useful. If anyone else experiences this problem, please
contact Sleepycat Support.

Regards,
Michael.