Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Nack, as we discussed earlier, please use new debug levels in the second patch.
Jan
On Tue, Nov 01, 2011 at 04:13:20PM +0100, Jan Zelený wrote:
Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Nack, as we discussed earlier, please use new debug levels in the second patch.
Jan
Done.
On Tue, Nov 01, 2011 at 04:13:20PM +0100, Jan Zelený wrote:
Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Nack, as we discussed earlier, please use new debug levels in the second patch.
Jan
Done.
Ack to both
Jan
On Tue, Nov 01, 2011 at 04:37:38PM +0100, Jan Zelený wrote:
On Tue, Nov 01, 2011 at 04:13:20PM +0100, Jan Zelený wrote:
Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Nack, as we discussed earlier, please use new debug levels in the second patch.
Jan
Done.
Ack to both
Jan
I'd like to request that the first patch is pushed to the 1.6 branch as well. Simo said he managed to hit the segfault himself, so we might as well release the fix in the 1.6.3 release.
On Wed, 2011-11-02 at 10:08 +0100, Jakub Hrozek wrote:
On Tue, Nov 01, 2011 at 04:37:38PM +0100, Jan Zelený wrote:
On Tue, Nov 01, 2011 at 04:13:20PM +0100, Jan Zelený wrote:
Simo was studying the dereference code lately and found two issues. #1 is an important bug, #2 is more of a cosmetic issue.
[PATCH 1/2] Use LDAPDerefSpec properly ldap_create_deref_control_value expects an array of LDAPDerefSpec structures with LDAPDerefSpec.derefAttr == NULL as a sentinel. We were passing a single instance of a LDAPDerefSpec structure.
https://fedorahosted.org/sssd/ticket/1050
[PATCH 2/2] Remove confusing do-while loop The deref processing would return a single control back. The do-while loop was harmless but confusing.
Nack, as we discussed earlier, please use new debug levels in the second patch.
Jan
Done.
Ack to both
Jan
I'd like to request that the first patch is pushed to the 1.6 branch as well. Simo said he managed to hit the segfault himself, so we might as well release the fix in the 1.6.3 release.
Both patches pushed to master, and the first patch pushed to sssd-1-6.
sssd-devel@lists.fedorahosted.org