提交 0838aa7f 编写于 作者: P Pablo Neira Ayuso

netfilter: fix netns dependencies with conntrack templates

Quoting Daniel Borkmann:

"When adding connection tracking template rules to a netns, f.e. to
configure netfilter zones, the kernel will endlessly busy-loop as soon
as we try to delete the given netns in case there's at least one
template present, which is problematic i.e. if there is such bravery that
the priviledged user inside the netns is assumed untrusted.

Minimal example:

  ip netns add foo
  ip netns exec foo iptables -t raw -A PREROUTING -d 1.2.3.4 -j CT --zone 1
  ip netns del foo

What happens is that when nf_ct_iterate_cleanup() is being called from
nf_conntrack_cleanup_net_list() for a provided netns, we always end up
with a net->ct.count > 0 and thus jump back to i_see_dead_people. We
don't get a soft-lockup as we still have a schedule() point, but the
serving CPU spins on 100% from that point onwards.

Since templates are normally allocated with nf_conntrack_alloc(), we
also bump net->ct.count. The issue why they are not yet nf_ct_put() is
because the per netns .exit() handler from x_tables (which would eventually
invoke xt_CT's xt_ct_tg_destroy() that drops reference on info->ct) is
called in the dependency chain at a *later* point in time than the per
netns .exit() handler for the connection tracker.

This is clearly a chicken'n'egg problem: after the connection tracker
.exit() handler, we've teared down all the connection tracking
infrastructure already, so rightfully, xt_ct_tg_destroy() cannot be
invoked at a later point in time during the netns cleanup, as that would
lead to a use-after-free. At the same time, we cannot make x_tables depend
on the connection tracker module, so that the xt_ct_tg_destroy() would
be invoked earlier in the cleanup chain."

Daniel confirms this has to do with the order in which modules are loaded or
having compiled nf_conntrack as modules while x_tables built-in. So we have no
guarantees regarding the order in which netns callbacks are executed.

Fix this by allocating the templates through kmalloc() from the respective
SYNPROXY and CT targets, so they don't depend on the conntrack kmem cache.
Then, release then via nf_ct_tmpl_free() from destroy_conntrack(). This branch
is marked as unlikely since conntrack templates are rarely allocated and only
from the configuration plane path.

Note that templates are not kept in any list to avoid further dependencies with
nf_conntrack anymore, thus, the tmpl larval list is removed.
Reported-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>
Tested-by: NDaniel Borkmann <daniel@iogearbox.net>
上级 484836ec
......@@ -291,7 +291,7 @@ extern unsigned int nf_conntrack_max;
extern unsigned int nf_conntrack_hash_rnd;
void init_nf_conntrack_hash_rnd(void);
void nf_conntrack_tmpl_insert(struct net *net, struct nf_conn *tmpl);
struct nf_conn *nf_ct_tmpl_alloc(struct net *net, u16 zone, gfp_t flags);
#define NF_CT_STAT_INC(net, count) __this_cpu_inc((net)->ct.stat->count)
#define NF_CT_STAT_INC_ATOMIC(net, count) this_cpu_inc((net)->ct.stat->count)
......
......@@ -68,7 +68,6 @@ struct ct_pcpu {
spinlock_t lock;
struct hlist_nulls_head unconfirmed;
struct hlist_nulls_head dying;
struct hlist_nulls_head tmpl;
};
struct netns_ct {
......
......@@ -287,6 +287,46 @@ static void nf_ct_del_from_dying_or_unconfirmed_list(struct nf_conn *ct)
spin_unlock(&pcpu->lock);
}
/* Released via destroy_conntrack() */
struct nf_conn *nf_ct_tmpl_alloc(struct net *net, u16 zone, gfp_t flags)
{
struct nf_conn *tmpl;
tmpl = kzalloc(sizeof(struct nf_conn), GFP_KERNEL);
if (tmpl == NULL)
return NULL;
tmpl->status = IPS_TEMPLATE;
write_pnet(&tmpl->ct_net, net);
#ifdef CONFIG_NF_CONNTRACK_ZONES
if (zone) {
struct nf_conntrack_zone *nf_ct_zone;
nf_ct_zone = nf_ct_ext_add(tmpl, NF_CT_EXT_ZONE, GFP_ATOMIC);
if (!nf_ct_zone)
goto out_free;
nf_ct_zone->id = zone;
}
#endif
atomic_set(&tmpl->ct_general.use, 0);
return tmpl;
#ifdef CONFIG_NF_CONNTRACK_ZONES
out_free:
kfree(tmpl);
return NULL;
#endif
}
EXPORT_SYMBOL_GPL(nf_ct_tmpl_alloc);
static void nf_ct_tmpl_free(struct nf_conn *tmpl)
{
nf_ct_ext_destroy(tmpl);
nf_ct_ext_free(tmpl);
kfree(tmpl);
}
static void
destroy_conntrack(struct nf_conntrack *nfct)
{
......@@ -298,6 +338,10 @@ destroy_conntrack(struct nf_conntrack *nfct)
NF_CT_ASSERT(atomic_read(&nfct->use) == 0);
NF_CT_ASSERT(!timer_pending(&ct->timeout));
if (unlikely(nf_ct_is_template(ct))) {
nf_ct_tmpl_free(ct);
return;
}
rcu_read_lock();
l4proto = __nf_ct_l4proto_find(nf_ct_l3num(ct), nf_ct_protonum(ct));
if (l4proto && l4proto->destroy)
......@@ -540,28 +584,6 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)
}
EXPORT_SYMBOL_GPL(nf_conntrack_hash_check_insert);
/* deletion from this larval template list happens via nf_ct_put() */
void nf_conntrack_tmpl_insert(struct net *net, struct nf_conn *tmpl)
{
struct ct_pcpu *pcpu;
__set_bit(IPS_TEMPLATE_BIT, &tmpl->status);
__set_bit(IPS_CONFIRMED_BIT, &tmpl->status);
nf_conntrack_get(&tmpl->ct_general);
/* add this conntrack to the (per cpu) tmpl list */
local_bh_disable();
tmpl->cpu = smp_processor_id();
pcpu = per_cpu_ptr(nf_ct_net(tmpl)->ct.pcpu_lists, tmpl->cpu);
spin_lock(&pcpu->lock);
/* Overload tuple linked list to put us in template list. */
hlist_nulls_add_head_rcu(&tmpl->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
&pcpu->tmpl);
spin_unlock_bh(&pcpu->lock);
}
EXPORT_SYMBOL_GPL(nf_conntrack_tmpl_insert);
/* Confirm a connection given skb; places it in hash table */
int
__nf_conntrack_confirm(struct sk_buff *skb)
......@@ -1751,7 +1773,6 @@ int nf_conntrack_init_net(struct net *net)
spin_lock_init(&pcpu->lock);
INIT_HLIST_NULLS_HEAD(&pcpu->unconfirmed, UNCONFIRMED_NULLS_VAL);
INIT_HLIST_NULLS_HEAD(&pcpu->dying, DYING_NULLS_VAL);
INIT_HLIST_NULLS_HEAD(&pcpu->tmpl, TEMPLATE_NULLS_VAL);
}
net->ct.stat = alloc_percpu(struct ip_conntrack_stat);
......
......@@ -349,12 +349,10 @@ static void __net_exit synproxy_proc_exit(struct net *net)
static int __net_init synproxy_net_init(struct net *net)
{
struct synproxy_net *snet = synproxy_pernet(net);
struct nf_conntrack_tuple t;
struct nf_conn *ct;
int err = -ENOMEM;
memset(&t, 0, sizeof(t));
ct = nf_conntrack_alloc(net, 0, &t, &t, GFP_KERNEL);
ct = nf_ct_tmpl_alloc(net, 0, GFP_KERNEL);
if (IS_ERR(ct)) {
err = PTR_ERR(ct);
goto err1;
......@@ -365,7 +363,8 @@ static int __net_init synproxy_net_init(struct net *net)
if (!nfct_synproxy_ext_add(ct))
goto err2;
nf_conntrack_tmpl_insert(net, ct);
__set_bit(IPS_CONFIRMED_BIT, &ct->status);
nf_conntrack_get(&ct->ct_general);
snet->tmpl = ct;
snet->stats = alloc_percpu(struct synproxy_stats);
......
......@@ -184,7 +184,6 @@ xt_ct_set_timeout(struct nf_conn *ct, const struct xt_tgchk_param *par,
static int xt_ct_tg_check(const struct xt_tgchk_param *par,
struct xt_ct_target_info_v1 *info)
{
struct nf_conntrack_tuple t;
struct nf_conn *ct;
int ret = -EOPNOTSUPP;
......@@ -202,8 +201,7 @@ static int xt_ct_tg_check(const struct xt_tgchk_param *par,
if (ret < 0)
goto err1;
memset(&t, 0, sizeof(t));
ct = nf_conntrack_alloc(par->net, info->zone, &t, &t, GFP_KERNEL);
ct = nf_ct_tmpl_alloc(par->net, info->zone, GFP_KERNEL);
ret = PTR_ERR(ct);
if (IS_ERR(ct))
goto err2;
......@@ -227,8 +225,8 @@ static int xt_ct_tg_check(const struct xt_tgchk_param *par,
if (ret < 0)
goto err3;
}
nf_conntrack_tmpl_insert(par->net, ct);
__set_bit(IPS_CONFIRMED_BIT, &ct->status);
nf_conntrack_get(&ct->ct_general);
out:
info->ct = ct;
return 0;
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册