On (30/08/16 17:07), Lukas Slebodnik wrote: >On (30/08/16 16:59), Fabiano Fidêncio wrote: >>Lukaš, >> >>On Tue, Aug 30, 2016 at 4:54 PM, Lukas Slebodnik <[email protected]> wrote: >>> ehlo, >>> >>> Clang static analyzer assume that ldb_search can found >>> 0 entries in the tree "cn=sysdb". Thenvariable version >>> could be used uninitialized. >>> >>> We cannot get to such state in sssd but we already handle >>> a case for more then one entry. >> >>I don't think this is the right approach as res->count == 0 seems to >>be a valid case for a newly created database (please, correct me if >>I'm wrong). >> >Agree > >I should have tried to run unit test before sending a patch > I looked deeper to the clang report And there are wrong assumption that output variable "version" is not initialized if function sysdb_cache_connect returns ERR_SYSDB_VERSION_TOO_OLD or ERR_SYSDB_VERSION_TOO_NEW
The reality is that output variable "version" is initialized especially for these two case. It is a false positive but we might suppress the warning with initializing variable to NULL. Attached is a gzipped html report from clang LS
report-f718a0.html.gz
Description: application/gzip
_______________________________________________ sssd-devel mailing list [email protected] https://lists.fedorahosted.org/admin/lists/[email protected]
