segfaulting cache manager (ITS#207)

[gene@imsa.edu]

Full_Name: Gene Skonicki
Version: 1.2.3
OS: Redhat Linux 6.0
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (143.195.10.100)


> index_add_values( "gecos", 2 )
=> index_add_values( "userpassword", 2 )
=> id2entry_add( 2, "uid=vader21,dc=imsa,dc=edu" )
=> ldbm_cache_open( "/usr/tmp/id2entry.dbb", 7, 600 )
<= ldbm_cache_open (cache 1)
<= id2entry_add 0
send_ldap_result 0::
ber_flush: 14 bytes to sd 8
         0 0c 02 01 02  g 07 0a 01 00 04 00 04 00
====> cache_return_entry_w
entry_rdwr_wunlock: ID: 2

The above is debugging output consistent
with the sorts of failures we see.   I cannot reproduce the problem with
complete regularity, however.  It only seems to happen on some ldap modifies. 
However, the end of things always looks roughly like that.  A gdb backtrace
shows me only that it is entering a thread and does a set of strcpy()'s which
finally result in a segfault.