nss-pam-ldapd before 0.7.18 and 0.8.x before 0.8.11 allows context-dependent attackers to cause a denial of service (application crash) and possibly execute arbitrary code by performing a name lookup on an application with a large number of open file descriptors, which triggers a stack-based buffer overflow related to incorrect use of the FD_SET macro.

nslcd/pam.c in the nss-pam-ldapd 0.8.0 PAM module returns a success code when a user is not found in LDAP, which allows remote attackers to bypass authentication.

Unknown vulnerability in pam_ldap before 180 does not properly handle a new password policy control, which could allow attackers to gain privileges. NOTE: CVE-2005-2497 had also been assigned to this issue, but CVE-2005-2641 is the correct candidate.

pam_ldap and nss_ldap, when used with OpenLDAP and connecting to a slave using TLS, does not use TLS for the subsequent connection if the client is referred to a master, which may cause a password to be sent in cleartext and allows remote attackers to sniff the password.

Unknown vulnerability in the pam_filter mechanism in pam_ldap before version 162, when LDAP based authentication is being used, allows users to bypass host-based access restrictions and log onto the system.

Format string vulnerability in the logging() function in C-Note Squid LDAP authentication module (squid_auth_LDAP) 2.0.2 and earlier allows remote attackers to cause a denial of service and possibly execute arbitrary code by triggering log messages.

Format string vulnerability in the logging function for the pam_ldap PAM LDAP module before version 144 allows attackers to execute arbitrary code via format strings in the configuration file name.