5.10-stable review patch. If anyone has any objections, please let me know. ------------------ From: Jozsef Kadlecsik <kadlec@xxxxxxxxxxxxx> [ Upstream commit 8ecd06277a7664f4ef018abae3abd3451d64e7a6 ] When destroying all sets, we are either in pernet exit phase or are executing a "destroy all sets command" from userspace. The latter was taken into account in ip_set_dereference() (nfnetlink mutex is held), but the former was not. The patch adds the required check to rcu_dereference_protected() in ip_set_dereference(). Fixes: 4e7aaa6b82d6 ("netfilter: ipset: Fix race between namespace cleanup and gc in the list:set type") Reported-by: syzbot+b62c37cdd58103293a5a@xxxxxxxxxxxxxxxxxxxxxxxxx Reported-by: syzbot+cfbe1da5fdfc39efc293@xxxxxxxxxxxxxxxxxxxxxxxxx Reported-by: kernel test robot <oliver.sang@xxxxxxxxx> Closes: https://lore.kernel.org/oe-lkp/202406141556.e0b6f17e-lkp@xxxxxxxxx Signed-off-by: Jozsef Kadlecsik <kadlec@xxxxxxxxxxxxx> Signed-off-by: Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx> --- net/netfilter/ipset/ip_set_core.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/net/netfilter/ipset/ip_set_core.c b/net/netfilter/ipset/ip_set_core.c index ecd693fca224f..bac92369a5436 100644 --- a/net/netfilter/ipset/ip_set_core.c +++ b/net/netfilter/ipset/ip_set_core.c @@ -53,12 +53,13 @@ MODULE_DESCRIPTION("core IP set support"); MODULE_ALIAS_NFNL_SUBSYS(NFNL_SUBSYS_IPSET); /* When the nfnl mutex or ip_set_ref_lock is held: */ -#define ip_set_dereference(p) \ - rcu_dereference_protected(p, \ +#define ip_set_dereference(inst) \ + rcu_dereference_protected((inst)->ip_set_list, \ lockdep_nfnl_is_held(NFNL_SUBSYS_IPSET) || \ - lockdep_is_held(&ip_set_ref_lock)) + lockdep_is_held(&ip_set_ref_lock) || \ + (inst)->is_deleted) #define ip_set(inst, id) \ - ip_set_dereference((inst)->ip_set_list)[id] + ip_set_dereference(inst)[id] #define ip_set_ref_netlink(inst,id) \ rcu_dereference_raw((inst)->ip_set_list)[id] #define ip_set_dereference_nfnl(p) \ @@ -1137,7 +1138,7 @@ static int ip_set_create(struct net *net, struct sock *ctnl, if (!list) goto cleanup; /* nfnl mutex is held, both lists are valid */ - tmp = ip_set_dereference(inst->ip_set_list); + tmp = ip_set_dereference(inst); memcpy(list, tmp, sizeof(struct ip_set *) * inst->ip_set_max); rcu_assign_pointer(inst->ip_set_list, list); /* Make sure all current packets have passed through */ -- 2.43.0