Re: Race condition on change of expired password?
Russ Allbery <[email protected]>
| Newsgroups | gmane.comp.encryption.kerberos.heimdal.general |
|---|---|
| Organization | The Eyrie |
| Message-ID | <[email protected]> |
Russ Allbery <[email protected]> writes: > This looks like a race condition, although I'm not exactly sure what to do > about it. > windlord:~/dvl/pam-krb5> kinit [email protected] > [email protected]'s Password: > Your password will expire at Wed Dec 21 17:41:10 2011 > Changing password > New password: > Repeat new password: > Success: Password changed > kinit: Password incorrect > windlord:~/dvl/pam-krb5> kinit [email protected] > [email protected]'s Password: > windlord:~/dvl/pam-krb5> echo $? > 0 It doesn't seem to *just* be a race condition. I redid this test inside the Kerberos PAM module and added a forced five second delay in the prompter after printing the success message, but the subsequent AS-REQ still fails: Dec 21 18:00:22 kerberos-dev kpasswdd[946]: Changing password for [email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: AS-REQ [email protected] from IPv4:171.67.225.134 for krbtgt/[email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: No preauth found, returning PREAUTH-REQUIRED -- [email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: sending 303 bytes to IPv4:171.67.225.134 Dec 21 18:00:27 kerberos-dev kdc[944]: AS-REQ [email protected] from IPv4:171.67.225.134 for krbtgt/[email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: Client sent patypes: encrypted-timestamp Dec 21 18:00:27 kerberos-dev kdc[944]: Looking for PKINIT pa-data -- [email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: Looking for ENC-TS pa-data -- [email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: Failed to decrypt PA-DATA -- [email protected] (enctype aes256-cts-hmac-sha1-96) error Decrypt integrity check failed Dec 21 18:00:27 kerberos-dev kdc[944]: Failed to decrypt PA-DATA -- [email protected] Dec 21 18:00:27 kerberos-dev kdc[944]: sending 159 bytes to IPv4:171.67.225.134 You can see the five second delay. So something else must be going on. -- Russ Allbery ([email protected]) <http://www.eyrie.org/~eagle/>