On Mon 24 Sep 2012 11:24:59 AM EDT, Stephen Gallagher wrote:
On Mon 24 Sep 2012 11:01:20 AM EDT, Pavel Březina wrote:
Recent AD patches introduced another warning.
If I read the code correctly it is safe to just initialize the variable to NULL.
Nack, this would not be safe. We would then be attempting to talloc_strdup() the NULL value at line 367. What is actually wrong here is that the error check from sysdb_search_group_by_gid() needs to fail if it returns anything but EOK or ENOENT, which I forgot to include. Falling through below is bad.
Thanks for catching that.
Whoops, forgot to attach the patch.
On Mon, Sep 24, 2012 at 11:32:52AM -0400, Stephen Gallagher wrote:
On Mon 24 Sep 2012 11:24:59 AM EDT, Stephen Gallagher wrote:
On Mon 24 Sep 2012 11:01:20 AM EDT, Pavel Březina wrote:
Recent AD patches introduced another warning.
If I read the code correctly it is safe to just initialize the variable to NULL.
Nack, this would not be safe. We would then be attempting to talloc_strdup() the NULL value at line 367. What is actually wrong here is that the error check from sysdb_search_group_by_gid() needs to fail if it returns anything but EOK or ENOENT, which I forgot to include. Falling through below is bad.
Thanks for catching that.
Whoops, forgot to attach the patch.
Ack and pushed to master.
sssd-devel@lists.fedorahosted.org