[krbdev.mit.edu #8331] potential memleak of pol_entry->name in populate_policy()

[Greg Hudson via RT]

It's a little more complicated than that.  pol_entry->name isn't strictly 
leaked since it's still accessible to the caller--although by current 
practices, a function like that shouldn't leave behind a partly-populated 
structure for the caller to clean up on error.

Of the two callers of that function, 
krb5_ldap_get_password_policy_from_dn() cleans up the policy structure on 
error, so can't leak memory if I'm reading it correctly.  
krb5_ldap_iterate_password_policy() does not appear to clean up the 
policy structure on error, so can leak memory.