Hi,
We have a long running server that tries to get a new LDAP handle if the
current handle becomes invalid (LDAP_SERVER_DOWN error, etc).
However if I call ldap_unbind_ext_s() on the old handle to release it, it
hangs with the following stack:
(gdb) bt
#0 0x000000080166935c in read () at read.S:2
#1 0x00000008014748f0 in __read (fd=68, buf=0x803293003, nbytes=5) at
/.automount/nfs.panwest.panasas.com/root/sb10/hlewis/hlewis-sb10-pioneer-ldap/src/freebsd-c/lib/libthr/thread/thr_syscalls.c:384
#2 0x0000000802c715d2 in sb_stream_read (sbiod=0x8088070d0, buf=0x803293003,
len=5) at
/.automount/nfs.panwest.panasas.com/root/sb10/hlewis/hlewis-sb10-pioneer-ldap/src/likewise/openldap/libraries/liblber/sockbuf.c:493
#3 0x0000000802c7207e in sb_debug_read (sbiod=0x803291220, buf=0x803293003,
len=5) at
/.automount/nfs.panwest.panasas.com/root/sb10/hlewis/hlewis-sb10-pioneer-ldap/src/likewise/openldap/libraries/liblber/sockbuf.c:827
#4 0x0000000802356971 in tlso_bio_read (b=0x803205400, buf=0x803293003
"\027\003\001", len=5) at tls_o.c:687
#5 0x000000080252c2df in BIO_read () from /usr/lib/libcrypto.so.1.0.0
#6 0x0000000802eb9f01 in ssl3_read_n () from /usr/lib/libssl.so.1.0.0
#7 0x0000000802eba3ce in ssl3_read_bytes () from /usr/lib/libssl.so.1.0.0
#8 0x0000000802eb67ac in ssl3_shutdown () from /usr/lib/libssl.so.1.0.0
#9 0x0000000802356cf6 in tlso_sb_close (sbiod=0x803291250) at tls_o.c:808
#10 0x0000000802c71253 in ber_int_sb_close (sb=0x803291160) at
/.automount/nfs.panwest.panasas.com/root/sb10/hlewis/hlewis-sb10-pioneer-ldap/src/likewise/openldap/libraries/liblber/sockbuf.c:383
#11 0x000000080233d625 in ldap_free_connection (ld=0x804c48c00,
lc=0x8088080f0, force=1, unbind=1) at request.c:769
#12 0x000000080232ffa2 in ldap_ld_free (ld=0x804c48c00, close=1, sctrls=0x0,
cctrls=0x0) at unbind.c:93
#13 0x000000080232fe8b in ldap_unbind_ext (ld=0x804c48c00, sctrls=0x0,
cctrls=0x0) at unbind.c:52
#14 0x000000080232fec5 in ldap_unbind_ext_s (ld=0x804c48c00, sctrls=0x0,
cctrls=0x0) at unbind.c:61
……
I want to make sure that there are memory leaks etc, hence the attempt to
release the invalid handle.
Question: What is the best or correct way to release an invalid LDAP handle?