netfilter: nf_conntrack: don't release a conntrack with non-zero refcnt

With this patch, the conntrack refcount is initially set to zero and
it is bumped once it is added to any of the list, so we fulfill
Eric's golden rule which is that all released objects always have a
refcount that equals zero.

Andrey Vagin reports that nf_conntrack_free can't be called for a
conntrack with non-zero ref-counter, because it can race with
nf_conntrack_find_get().

A conntrack slab is created with SLAB_DESTROY_BY_RCU. Non-zero
ref-counter says that this conntrack is used. So when we release
a conntrack with non-zero counter, we break this assumption.

CPU1                                    CPU2
____nf_conntrack_find()
                                        nf_ct_put()
                                         destroy_conntrack()
                                        ...
                                        init_conntrack
                                         __nf_conntrack_alloc (set use = 1)
atomic_inc_not_zero(&ct->use) (use = 2)
                                         if (!l4proto->new(ct, skb, dataoff, timeouts))
                                          nf_conntrack_free(ct); (use = 2 !!!)
                                        ...
                                        __nf_conntrack_alloc (set use = 1)
 if (!nf_ct_key_equal(h, tuple, zone))
  nf_ct_put(ct); (use = 0)
   destroy_conntrack()
                                        /* continue to work with CT */

After applying the path "[PATCH] netfilter: nf_conntrack: fix RCU
race in nf_conntrack_find_get" another bug was triggered in
destroy_conntrack():

<4>[67096.759334] ------------[ cut here ]------------
<2>[67096.759353] kernel BUG at net/netfilter/nf_conntrack_core.c:211!
...
<4>[67096.759837] Pid: 498649, comm: atdd veid: 666 Tainted: G         C ---------------    2.6.32-042stab084.18 #1 042stab084_18 /DQ45CB
<4>[67096.759932] RIP: 0010:[<ffffffffa03d99ac>]  [<ffffffffa03d99ac>] destroy_conntrack+0x15c/0x190 [nf_conntrack]
<4>[67096.760255] Call Trace:
<4>[67096.760255]  [<ffffffff814844a7>] nf_conntrack_destroy+0x17/0x30
<4>[67096.760255]  [<ffffffffa03d9bb5>] nf_conntrack_find_get+0x85/0x130 [nf_conntrack]
<4>[67096.760255]  [<ffffffffa03d9fb2>] nf_conntrack_in+0x352/0xb60 [nf_conntrack]
<4>[67096.760255]  [<ffffffffa048c771>] ipv4_conntrack_local+0x51/0x60 [nf_conntrack_ipv4]
<4>[67096.760255]  [<ffffffff81484419>] nf_iterate+0x69/0xb0
<4>[67096.760255]  [<ffffffff814b5b00>] ? dst_output+0x0/0x20
<4>[67096.760255]  [<ffffffff814845d4>] nf_hook_slow+0x74/0x110
<4>[67096.760255]  [<ffffffff814b5b00>] ? dst_output+0x0/0x20
<4>[67096.760255]  [<ffffffff814b66d5>] raw_sendmsg+0x775/0x910
<4>[67096.760255]  [<ffffffff8104c5a8>] ? flush_tlb_others_ipi+0x128/0x130
<4>[67096.760255]  [<ffffffff8100bc4e>] ? apic_timer_interrupt+0xe/0x20
<4>[67096.760255]  [<ffffffff8100bc4e>] ? apic_timer_interrupt+0xe/0x20
<4>[67096.760255]  [<ffffffff814c136a>] inet_sendmsg+0x4a/0xb0
<4>[67096.760255]  [<ffffffff81444e93>] ? sock_sendmsg+0x13/0x140
<4>[67096.760255]  [<ffffffff81444f97>] sock_sendmsg+0x117/0x140
<4>[67096.760255]  [<ffffffff8102e299>] ? native_smp_send_reschedule+0x49/0x60
<4>[67096.760255]  [<ffffffff81519beb>] ? _spin_unlock_bh+0x1b/0x20
<4>[67096.760255]  [<ffffffff8109d930>] ? autoremove_wake_function+0x0/0x40
<4>[67096.760255]  [<ffffffff814960f0>] ? do_ip_setsockopt+0x90/0xd80
<4>[67096.760255]  [<ffffffff8100bc4e>] ? apic_timer_interrupt+0xe/0x20
<4>[67096.760255]  [<ffffffff8100bc4e>] ? apic_timer_interrupt+0xe/0x20
<4>[67096.760255]  [<ffffffff814457c9>] sys_sendto+0x139/0x190
<4>[67096.760255]  [<ffffffff810efa77>] ? audit_syscall_entry+0x1d7/0x200
<4>[67096.760255]  [<ffffffff810ef7c5>] ? __audit_syscall_exit+0x265/0x290
<4>[67096.760255]  [<ffffffff81474daf>] compat_sys_socketcall+0x13f/0x210
<4>[67096.760255]  [<ffffffff8104dea3>] ia32_sysret+0x0/0x5

I have reused the original title for the RFC patch that Andrey posted and
most of the original patch description.

Cc: Eric Dumazet <edumazet@google.com>
Cc: Andrew Vagin <avagin@parallels.com>
Cc: Florian Westphal <fw@strlen.de>
Reported-by: Andrew Vagin <avagin@parallels.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Andrew Vagin <avagin@parallels.com>

diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
index 01ea6eed1bb1ddcc9b8c9001286ae85730bea416..b2ac6246b7e0abe156b26a06bf9a4463331baa6e 100644 (file)
--- a/include/net/netfilter/nf_conntrack.h
+++ b/include/net/netfilter/nf_conntrack.h
@@ -284,6 +284,8 @@ 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);
+
 #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)
 

diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 4d1fb5d094c3ba98f40870a71ea44b0fe8f37f03..356bef519fe5b781f6225557f64186c0075852a1 100644 (file)
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -448,7 +448,9 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)
                        goto out;
 
        add_timer(&ct->timeout);
-       nf_conntrack_get(&ct->ct_general);
+       smp_wmb();
+       /* The caller holds a reference to this object */
+       atomic_set(&ct->ct_general.use, 2);
        __nf_conntrack_hash_insert(ct, hash, repl_hash);
        NF_CT_STAT_INC(net, insert);
        spin_unlock_bh(&nf_conntrack_lock);
@@ -462,6 +464,21 @@ out:
 }
 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)
+{
+       __set_bit(IPS_TEMPLATE_BIT, &tmpl->status);
+       __set_bit(IPS_CONFIRMED_BIT, &tmpl->status);
+       nf_conntrack_get(&tmpl->ct_general);
+
+       spin_lock_bh(&nf_conntrack_lock);
+       /* Overload tuple linked list to put us in template list. */
+       hlist_nulls_add_head_rcu(&tmpl->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
+                                &net->ct.tmpl);
+       spin_unlock_bh(&nf_conntrack_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)
@@ -733,11 +750,10 @@ __nf_conntrack_alloc(struct net *net, u16 zone,
                nf_ct_zone->id = zone;
        }
 #endif
-       /*
-        * changes to lookup keys must be done before setting refcnt to 1
+       /* Because we use RCU lookups, we set ct_general.use to zero before
+        * this is inserted in any list.
         */
-       smp_wmb();
-       atomic_set(&ct->ct_general.use, 1);
+       atomic_set(&ct->ct_general.use, 0);
        return ct;
 
 #ifdef CONFIG_NF_CONNTRACK_ZONES
@@ -761,6 +777,11 @@ void nf_conntrack_free(struct nf_conn *ct)
 {
        struct net *net = nf_ct_net(ct);
 
+       /* A freed object has refcnt == 0, that's
+        * the golden rule for SLAB_DESTROY_BY_RCU
+        */
+       NF_CT_ASSERT(atomic_read(&ct->ct_general.use) == 0);
+
        nf_ct_ext_destroy(ct);
        nf_ct_ext_free(ct);
        kmem_cache_free(net->ct.nf_conntrack_cachep, ct);
@@ -856,6 +877,9 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,
                NF_CT_STAT_INC(net, new);
        }
 
+       /* Now it is inserted into the unconfirmed list, bump refcount */
+       nf_conntrack_get(&ct->ct_general);
+
        /* Overload tuple linked list to put us in unconfirmed list. */
        hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
                       &net->ct.unconfirmed);

diff --git a/net/netfilter/nf_synproxy_core.c b/net/netfilter/nf_synproxy_core.c
index 9858e3e51a3a049ce796b3ed625e3d3ad8bbe5cc..52e20c9a46a58be4328276964c1bb255c51a423c 100644 (file)
--- a/net/netfilter/nf_synproxy_core.c
+++ b/net/netfilter/nf_synproxy_core.c
@@ -363,9 +363,8 @@ static int __net_init synproxy_net_init(struct net *net)
                goto err2;
        if (!nfct_synproxy_ext_add(ct))
                goto err2;
-       __set_bit(IPS_TEMPLATE_BIT, &ct->status);
-       __set_bit(IPS_CONFIRMED_BIT, &ct->status);
 
+       nf_conntrack_tmpl_insert(net, ct);
        snet->tmpl = ct;
 
        snet->stats = alloc_percpu(struct synproxy_stats);
@@ -390,7 +389,7 @@ static void __net_exit synproxy_net_exit(struct net *net)
 {
        struct synproxy_net *snet = synproxy_pernet(net);
 
-       nf_conntrack_free(snet->tmpl);
+       nf_ct_put(snet->tmpl);
        synproxy_proc_exit(net);
        free_percpu(snet->stats);
 }

diff --git a/net/netfilter/xt_CT.c b/net/netfilter/xt_CT.c
index 5929be622c5cd27b9e706811042c9e176a3c78f7..75747aecdebe6344ccbfcf178c299be013d8f763 100644 (file)
--- a/net/netfilter/xt_CT.c
+++ b/net/netfilter/xt_CT.c
@@ -228,12 +228,7 @@ static int xt_ct_tg_check(const struct xt_tgchk_param *par,
                        goto err3;
        }
 
-       __set_bit(IPS_TEMPLATE_BIT, &ct->status);
-       __set_bit(IPS_CONFIRMED_BIT, &ct->status);
-
-       /* Overload tuple linked list to put us in template list. */
-       hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
-                                &par->net->ct.tmpl);
+       nf_conntrack_tmpl_insert(par->net, ct);
 out:
        info->ct = ct;
        return 0;