On Tue, Aug 18, 2020 at 9:06 AM Stephen Smalley
<stephen.smalley.work@xxxxxxxxx> wrote:
> Also, there appear to be other leaks of memory allocated near
> net/netlabel/netlabel_mgmt.c:88
> net/netlabel/netlabel_mgmt.c:201:
> 86 struct netlbl_dom_map *entry = kzalloc(sizeof(*entry),
> GFP_KERNEL);
> 87
> 88 if (!entry)
> 89 return -ENOMEM;
> ...
> 195 } else if (info->attrs[NLBL_MGMT_A_IPV6ADDR]) {
> 196 struct in6_addr *addr;
> 197 struct in6_addr *mask;
> 198 struct netlbl_domaddr6_map *map;
> 199
> 200 addrmap = kzalloc(sizeof(*addrmap), GFP_KERNEL);
> 201 if (addrmap == NULL) {
> 202 ret_val = -ENOMEM;
> 203 goto add_doi_put_def;
> 204 }
>
> Probably the same cause for all of them.
I've had a number of interruptions this week on a variety of fronts so
I've been late in looking in on this further, but at the very least it
looks like netlbl_domhsh_free_entry() needs to free
netlbl_dom_map->netlbl_dommap_def->addrsel in the
NETLBL_NLTYPE_ADDRSELECT case. I'll plug this hole and check the
rest.
--
paul moore
www.paul-moore.com
