On Wed, Jul 17, 2013 at 02:04:16PM +0200, Ondrej Kos wrote:
On 07/16/2013 08:10 PM, Jakub Hrozek wrote:
>On Tue, Jul 09, 2013 at 11:15:09PM +0200, Jakub Hrozek wrote:
>>On Fri, Jun 28, 2013 at 02:23:01PM +0200, Lukas Slebodnik wrote:
>>>On (24/06/13 17:04), Ondrej Kos wrote:
>>>>The problem here wasn't in returned error code, but in faultly read
>>>>DBUS message, due to condition in sss_authtok_set_string.
>>>>
>>>>When password is empty, it passes 0 as length, which is
>>>>misinterpreted, and the function tries to determine the length of
>>>>string by itself, reaching over boundaries of authtok string.
>>>>
>>>>trac issue:
https://fedorahosted.org/sssd/ticket/1814
>>>>
>>>>Patch is attached
>>>>
>>>>Ondra
>>>
>>>We found out that it is right approach (after long discussion :-)
>>>
>>>ACK
>>
>>The approach might be good but on entering a blank password I'm getting
>>"System Error" with this patch, while without the patch I was getting
>>"Authentication failure". PAM_AUTH_ERR is the correct return code for
>>this use case.
>
>Hi,
>
>Ondra asked me to re-test atop the current git HEAD. Here is what I see
>when I type "su - jhrozek" and then just type "Enter":
>
>[sssd[be[redhat.com]]] [be_pam_handler] (0x0100): Got request with the following
data
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): command: PAM_AUTHENTICATE
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): domain:
redhat.com
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): user: jhrozek
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): service: su-l
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): tty: pts/20
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): ruser: jhrozek
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): rhost:
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): authtok type: 0
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): newauthtok type: 0
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): priv: 0
>[sssd[be[redhat.com]]] [pam_print_data] (0x0100): cli_pid: 22305
>[sssd[be[redhat.com]]] [krb5_pam_handler] (0x1000): Wait queue of user [jhrozek] is
empty, running request immediately.
>[sssd[be[redhat.com]]] [krb5_auth_send] (0x0020): Wrong authtok type for user
[jhrozek]. Expected [1], got [0]
>[sssd[be[redhat.com]]] [check_wait_queue] (0x1000): Wait queue for user [jhrozek] is
empty.
>[sssd[be[redhat.com]]] [be_pam_handler_callback] (0x0100): Backend returned: (0, 4,
<NULL>) [Success]
>[sssd[be[redhat.com]]] [be_pam_handler_callback] (0x0100): Sending result
[
4][redhat.com]
>[sssd[be[redhat.com]]] [be_pam_handler_callback] (0x0100): Sent result
[
4][redhat.com]
>
>
>4 == System Error
>_______________________________________________
>sssd-devel mailing list
>sssd-devel(a)lists.fedorahosted.org
>https://lists.fedorahosted.org/mailman/listinfo/sssd-devel
>
Hi,
Now I see what you meant, new patches rebased on top of current master are
attached.
The third patch replaces mutiple calls of sss_authtok_get_type with variable
and single call, just a small refactoring.
Ondra
--
Ondrej Kos
Associate Software Engineer
Identity Management - SSSD
Red Hat Czech
[PATCH 1/3] Do not try to set password when authtok_length is zero
Ack
From 3968df17f835a8dac892b3756f2a4f35a86a0baa Mon Sep 17 00:00:00
2001
From: Ondrej Kos <okos(a)redhat.com>
Date: Wed, 17 Jul 2013 13:42:57 +0200
Subject: [PATCH 2/3] KRB: Handle empty password gracefully
https://fedorahosted.org/sssd/ticket/1814
Return authentication error when empty password is passed.
---
src/providers/krb5/krb5_auth.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/src/providers/krb5/krb5_auth.c b/src/providers/krb5/krb5_auth.c
index 22495f57079d979b6c3484acf916ecc0736292d1..5b4d4abb6c9f97cac394853d73eb4e5f8b045c44
100644
--- a/src/providers/krb5/krb5_auth.c
+++ b/src/providers/krb5/krb5_auth.c
@@ -495,6 +495,19 @@ struct tevent_req *krb5_auth_send(TALLOC_CTX *mem_ctx,
case SSS_PAM_AUTHENTICATE:
case SSS_PAM_CHAUTHTOK:
if (sss_authtok_get_type(pd->authtok) != SSS_AUTHTOK_TYPE_PASSWORD) {
+ /* handle empty password gracefully */
+ if (sss_authtok_get_type(pd->authtok) == SSS_AUTHTOK_TYPE_EMPTY
&&
+ sss_authtok_get_size(pd->authtok) == 0) {
The second part is not needed, empty authtok cannot have size != 0 by
definiton.
+
+ DEBUG(SSSDBG_CRIT_FAILURE,
+ ("Illegal zero-length authtok for user [%s]\n",
+ pd->user));
+ state->pam_status = PAM_AUTH_ERR;
+ state->dp_err = DP_ERR_OK;
+ ret = EOK;
+ goto done;
+ }
+
DEBUG(SSSDBG_CRIT_FAILURE,
("Wrong authtok type for user [%s]. " \
"Expected [%d], got [%d]\n", pd->user,
--
1.8.1.4
From e9f44ae4e487b0cddfb676c7f52bcef9780c4393 Mon Sep 17 00:00:00
2001
From: Ondrej Kos <okos(a)redhat.com>
Date: Wed, 17 Jul 2013 13:58:36 +0200
Subject: [PATCH 3/3] KRB: Replace multiple calls with variable
Meh, the compiler would optimize the call anyway :-) But in general I'm
not opposed.