Resolved deadlock by passing target entry to be_group and using
this if dn same as bdn.  It might actually be safer to check
entry ids instead of dns.
Resolved bogus add to cache after failed acl check by deferring
cache add until after parent/acl checks have successful been
completed.
Eliminated race condition caused by concurrent adds
of same dn by adding 'li_add_mutex' around the critical section
of code (most of ldbm_back_add).
This code is preliminary and still needs significant testing.

Seems that calling pthread_exit() in the main (and only active)
thread does not cause the whole process to exit.  Very odd.
Anyways, as we want to whole process to exit, we should just exit
after joining with our other threads.
I've also removed dead code for detaching threads we join with.

mutex declaration should be moved from slapd/main.c to slapd/init.c
so we don't have ripple changes through slapd/tools.

Introduction of ldbm_datum_init().
Patch for non-NULL pointer test befor call to ldbm_datum_free().

add/merge as we now free strings agressively.  Improved debug
message to include name of missing required attribute and added
check for 'operational attributes'.  This check should be used
everywhere we need to test for operational attributes (add/modify).
Also, enabled schema checking for tests (and fixed resulting
problems by adjusting oc.conf).

make it to dn2entry_r().

Also, output UNKNOWN_PREFIX ('?') instead of writing a nul character
when the attribute prefix is unknown.

to after other suffix checks.

Could still deadlock on the close().  A surrogate parent
should be used to create the children...

in ITS#17.

Updated other codes as needed.

call has been removed and additional code added to properly free
realBase variable.  Also fixed leaking of the matched variable.

	struct friendly (Friendly) changed to ldap_friendly (LDAPFriendly)
	  ldap_friendly friend prefix changed to 'lf_' from 'f_'
	removed mod_next field from LDAPMod (struct ldapmod)
	modified slapd to use new LDAPModList (struct ldapmodlist) struct.
Added LDAPv3 result codes to ldap.h

LDAP_SYSLOG in slap.h/slurp.h.

type 'sig_atomic_t' and marked it volatile.  Add detection to
configure to ensure sig_atomic_t and volatile are supported.

  Daniel Carroll <dan@mesa7.mesa.colorado.edu>
This patch ensures we try deleting the entry from both the
dntree and the idtree.

If Berkeley DB version 2.x is used as a slapd back-end a not-initialized
Datum structure will raise an EINVAL error in (DB)->put().

incorrect because value_cmp() in value.c compares two attributes by
comparing the leading bytes of the two values (candidate and filter) up to
the smaller of the two lengths of the binary values. If the lengths of two
SYNTAX_BIN values differ, like SYNTAX_CES and SYNTAX_CIS, the EQUALITY
comparison should fail.

It also appears that value_ncmp() in value.c has no references anywhere in
the openLDAP source tree, save for a definition in proto-slap.h. The
following fix the bug and remove the dead code:

From: "MCGARRAHAN, KENNETH L" <km4155@momail.sbc.com>

  "Check that the entry has no writers before removing it from the
  cache" sounds more sensible to me than "check that it *has* writers"
  before removing it.
I've also changed the default cachesize for tests to 4.  This
should help discover such problems.

Still leaks ber_scanf results here and there...