FYI: 2.6.13.3 Memory leak, names_cache

[Chris Wright]

* Steve Grubb (sgrubb at redhat.com) wrote:
> On Friday 07 October 2005 12:49, Timothy R. Chavez wrote:
> > Shouldn't this type of message be CCed to this list?
> 
> Offhand, I don't see the correlation between names_cache and audit system.

Following symlinks grabs memory from names_cache with raw __getname
meaning name is not added to names list in audit context.  It then uses
putname to put it back which leaks because __putname is postponed until
exit, and during exit audit_free_names doesn't find anything on the list.