Some newbie questions

["West, Michael, Jr (Mike)" <mjwest@avaya.com>]

Hopefully I can get some insight into what stupidity I am performing
here.

I have a very simple application that will do LDAP lookups on an LDAP
database using the C language libraries provided in OpenLDAP 2.2.13.
This program is compiled and run on RedHat version 9, but I have
replaced the OpenLDAP, SSL, SASL, and Berkeley DB libraries. While
the C language libraries are used, the actual program itself is
written in C++.

I tried to model the structure of the program based on the ldapsearch
utility, and the available man pages (which are just a tad antiquated).

The application is intended to do a sequence of lookups. There is no
set number; it will run until the user gets tired of making requests
and ends the program. (Or, in my case, until the program crashes.)

The sequence of calls I make is:
- ldap_initialize()
- ldap_set_option() (to version 3)
- ldap_simple_bind_s() (null password; directory requires none)
- ldap_search_ext_s()
- ldap_unbind_s()

(I intentionally use ldap_simple_bind_s and ldap_unbind_s despite
their being deprecated because (1) ldapsearch does and (2) I have
no idea what control lists are, and didn't think I needed to use
them.)

The actual search itself is very simple. I am simply providing
a single search key (email address) and returning a single value
(telephone number).

Initially, I was doing that full process for each request (from
the initialize through the unbind) in order to isolate each
request from the other. But when I did that, the program would
get a segmentation fault on the second call to ldap_simple_bind_s.

I then modified the program to call ldap_initialize, ldap_set_option,
and ldap_simple_bind_s once in the beginning, ldap_unbind_s once
at the end, and ldap_search_ext_s as many times as needed in
between. This time a segmentation fault would occur in different
spots, at after a variable number of times. The location was
usually in ldab_simple_bind_s, ldap_search_ext_s, or ldap_unbind_s.
The number of queries successfully completed ranged from 3 to 7.

In both versions, I ran the program under gdb and did a "back
trace". Regardless of the library function the segmentation fault
occurred in, the function causing the fault was a call to malloc().

This program seems like it should be very simple to create and
run. I have followed the above mentioned examples as closely as
I understand, but am obviously missing something. (I can provide
more detail on my code, but didn't want to clutter up this already
overlong message).

Thanks in advance for any merciful help.

Mike West
mjwest@avaya.com