[SSSD] [PATCH] Handle Krb5 password expiration warning

Stephen Gallagher sgallagh at redhat.com
Tue May 4 19:16:12 UTC 2010


On 05/03/2010 04:44 PM, Sumit Bose wrote:
> On Mon, May 03, 2010 at 10:39:57PM +0200, Sumit Bose wrote:
>> On Mon, May 03, 2010 at 02:29:39PM -0400, Stephen Gallagher wrote:
>>> On 04/30/2010 12:10 PM, Sumit Bose wrote:
>>>> On Fri, Apr 30, 2010 at 10:18:39AM -0400, Stephen Gallagher wrote:
>>>>> On 04/29/2010 07:27 AM, Sumit Bose wrote:
>>>>>> Sorry, the first patch had dependencies to another patch, new version
>>>>>> attached.
>>>>>>
>>>>>> bye,
>>>>>> Sumit
>>>>>> On Tue, Apr 27, 2010 at 02:08:02PM +0200, Sumit Bose wrote:
>>>>>>> Hi,
>>>>>>>
>>>>>>> this two patches add the support to display a warning to the user that
>>>>>>> the Kerberos password is about to expire. The first patch just moves
>>>>>>> some utility functions to a separate file to avoid linking the
>>>>>>> krb5_child against libdbus.
>>>>>>>
>>>>>>> I the second patch a prompter function is introduced to catch the
>>>>>>> warning message which is generated by libkrb5. With the current API of
>>>>>>> MIT Kerberos we have to rely on this message, because the the underlying
>>>>>>> AS_REPLY data is not exposed by the library. As a consequece this message
>>>>>>> cannot be translated, but this is true for messages which are generated
>>>>>>> to indicate why a password change failed, too. The rest of the patch
>>>>>>> changes the way the response from the child is packed and unpacked to
>>>>>>> allow more than one response message.
>>>>>>>
>>>>>>> bye,
>>>>>>> Sumit
>>>>>>
>>>>>
>>>>>
>>>>> Patch 0001: Ack
>>>>>
>>>>> Patch 0002: The krb5_child_done() function is in desperate need of
>>>>> comments. At minimum, there needs to be a comment describing the format
>>>>> of the received buffer. Same for the pack_response_packet().
>>>>
>>>> comments added
>>>>
>>>>>
>>>>> This is wrong:
>>>>>            if (msg_len>   pref_len&&
>>>>>                strncmp((const char *)&buf[p], CCACHE_ENV_NAME"=",
>>>>> pref_len) == 0) {
>>>>>
>>>>> pref_len needs to be another byte longer to account for the '='. Also, a
>>>>> comment here would be fantastic.
>>>>
>>>> pref_len is already a bye longer, because I used sizeof instead of
>>>> strlen. I added a comment about this, too.
>>>>
>>>> New version attached.
>>>>
>>>
>>> Sumit, these patches don't apply cleanly on sssd-1-2 at present. Would
>>> you please rebase them and resend?
>>>
>>
>> sure, rebased version attached.
>>
>
> ah, sorry, I have attached a broken version, this one should work.
>
> bye,
> Sumit


../../src/providers/krb5/krb5_auth.c: In function ‘krb5_save_ccname_done’:
../../src/providers/krb5/krb5_auth.c:1285: warning: unused variable 
‘krb5_ctx’

Otherwise, ack. I can fix this before I push.

-- 
Stephen Gallagher
RHCE 804006346421761

Delivering value year after year.
Red Hat ranks #1 in value among software vendors.
http://www.redhat.com/promo/vendor/



More information about the sssd-devel mailing list