[SSSD] [PATCHES] NSS: Clear cached netgroups if a request comes in from the sss_cache

Lukas Slebodnik lslebodn at redhat.com
Tue Aug 6 12:00:09 UTC 2013


On (06/08/13 13:52), Jakub Hrozek wrote:
>On Fri, Aug 02, 2013 at 06:44:51PM +0200, Lukas Slebodnik wrote:
>> On (31/07/13 12:59), Pavel Březina wrote:
>> >On 07/30/2013 08:19 AM, Lukas Slebodnik wrote:
>> >>ehlo,
>> >>
>> >>Attached patches fix ticket #1759
>> >>sss_cache -N/-n should invalidate the hash table in sssd_nss
>> >>
>> >>LS
>> >
>> >Nack.
>> >
>> >The hash table is cleared successfully, however the netgroup is not
>> >removed from the sysdb when it is not found in ldap. Therefore the
>> >following set of commands still returns the netgroup, although it
>> >should not:
>> >
>> >$ getent netgroup ng-1
>> >ng-1                  ( ,,bobby,example.com) ( ,,johny.example.com)
>> >...delete ng-1 from ldap
>> >$ sudo sss_cache -N
>> >$ getent netgroup ng-1
>> >ng-1                  ( ,,bobby,example.com) ( ,,johny.example.com)
>> >
>> It works with IPA, but I spent a lot of time with debugging this problem
>> and it is unrelated to my patches. Change(bug?) was introduced in commit
>> ca344fde. First patch should fix this
>
>Wow, nice catch. But I would expect a compiler warning or Coverity
>report here..this is a clear refactoring bug..
>
>Ack to the new patch. This needs to be pushed to Fedora right now.

I don't know why, but you can use NULL as a boolean argument.
There will be no warning even with -Wextra.
I check simple example with gcc, clang and clang static analyser.
It was a big surprise for me.

LS



More information about the sssd-devel mailing list