Netfilter fixes for net
Subsystems (3)
Pablo Neira Ayuso
maintainercover Netfilter fixes for net
Hi,
The following patchset contains Netfilter fixes for net:
1) refcount_inc_not_zero() is not semantically equivalent to atomic_int_not_zero(), from Florian Westphal. My understanding was that refcount_*() API provides a wrapper to easier debugging of reference count leaks, however, there are semantic differences between these two APIs, where refcount_inc_not_zero() needs a barrier. Reason for this subtle difference to me is unknown.
2) packet logging is not correct for ARP and IP packets, from the ARP family and netdev/egress respectively. Use skb_network_offset() to reach the headers accordingly.
3) set element extension length have been growing over time, replace a BUG_ON by EINVAL which might be triggerable from userspace.
Please, pull these changes from:
git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf.git
Thanks.
----------------------------------------------------------------
The following changes since commit 280e3a857d96f9ca8e24632788e1e7a0fec4e9f7:
Merge git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf (2022-07-03 12:29:18 +0100)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf.git HEAD
for you to fetch changes up to c39ba4de6b0a843bec5d46c2b6f2064428dada5e:
netfilter: nf_tables: replace BUG_ON by element length check (2022-07-09 16:25:09 +0200)
----------------------------------------------------------------
Florian Westphal (1):
netfilter: conntrack: fix crash due to confirmed bit load reordering
Pablo Neira Ayuso (2):
netfilter: nf_log: incorrect offset to network header
netfilter: nf_tables: replace BUG_ON by element length check
include/net/netfilter/nf_tables.h | 14 ++++--- net/netfilter/nf_conntrack_core.c | 22 ++++++++++ net/netfilter/nf_conntrack_netlink.c | 1 + net/netfilter/nf_conntrack_standalone.c | 3 ++ net/netfilter/nf_log_syslog.c | 8 ++-- net/netfilter/nf_tables_api.c | 72 +++++++++++++++++++++++---------- 6 files changed, 90 insertions(+), 30 deletions(-)
Pablo Neira Ayuso
maintainerPATCH 1/3 netfilter: conntrack: fix crash due to confirmed bit load reordering
From: Florian Westphal <[email protected]>
Kajetan Puchalski reports crash on ARM, with backtrace of:
__nf_ct_delete_from_listsnf_ct_deleteearly_drop__nf_conntrack_alloc
Unlike atomic_inc_not_zero, refcount_inc_not_zero is not a full barrier. conntrack uses SLAB_TYPESAFE_BY_RCU, i.e. it is possible that a 'newly' allocated object is still in use on another CPU:
CPU1 CPU2
encounter 'ct' during hlist walk
delete_from_lists
refcount drops to 0
kmem_cache_free(ct);
__nf_conntrack_alloc() // returns same object
refcount_inc_not_zero(ct); /* might fail */ /* If set, ct is public/in the hash table */ test_bit(IPS_CONFIRMED_BIT, &ct->status);
In case CPU1 already set refcount back to 1, refcount_inc_not_zero() will succeed.
The expected possibilities for a CPU that obtained the object 'ct' (but no reference so far) are:
1. refcount_inc_not_zero() fails. CPU2 ignores the object and moves to the next entry in the list. This happens for objects that are about to be free'd, that have been free'd, or that have been reallocated by __nf_conntrack_alloc(), but where the refcount has not been increased back to 1 yet.
2. refcount_inc_not_zero() succeeds. CPU2 checks the CONFIRMED bit in ct->status. If set, the object is public/in the table.
If not, the object must be skipped; CPU2 calls nf_ct_put() to un-do the refcount increment and moves to the next object.
Parallel deletion from the hlists is prevented by a 'test_and_set_bit(IPS_DYING_BIT, &ct->status);' check, i.e. only one cpu will do the unlink, the other one will only drop its reference count.
Because refcount_inc_not_zero is not a full barrier, CPU2 may try to delete an object that is not on any list:
1. refcount_inc_not_zero() successful (refcount inited to 1 on other CPU)
2. CONFIRMED test also successful (load was reordered or zeroing of ct->status not yet visible)
3. delete_from_lists unlinks entry not on the hlist, because IPS_DYING_BIT is 0 (already cleared).
2) is already wrong: CPU2 will handle a partially initited object that is supposed to be private to CPU1.
Add needed barriers when refcount_inc_not_zero() is successful.
It also inserts a smp_wmb() before the refcount is set to 1 during allocation.
Because other CPU might still see the object, refcount_set(1) "resurrects" it, so we need to make sure that other CPUs will also observe the right content. In particular, the CONFIRMED bit test must only pass once the object is fully initialised and either in the hash or about to be inserted (with locks held to delay possible unlink from early_drop or gc worker).
I did not change flow_offload_alloc(), as far as I can see it should call refcount_inc(), not refcount_inc_not_zero(): the ct object is attached to the skb so its refcount should be >= 1 in all cases.
v2: prefer smp_acquire__after_ctrl_dep to smp_rmb (Will Deacon).
v3: keep smp_acquire__after_ctrl_dep close to refcount_inc_not_zero call
add comment in nf_conntrack_netlink, no control dependency there due to locks.
--- net/netfilter/nf_conntrack_core.c | 22 ++++++++++++++++++++++ net/netfilter/nf_conntrack_netlink.c | 1 + net/netfilter/nf_conntrack_standalone.c | 3 +++ 3 files changed, 26 insertions(+)
net/netfilter/nf_conntrack_core.c +22 -0
@@ -729,6 +729,9 @@ static void nf_ct_gc_expired(struct nf_conn *ct)if (!refcount_inc_not_zero (&ct->ct_general.use))return ; +/* load ->status after refcount increase */ +smp_acquire__after_ctrl_dep (); +if (nf_ct_should_gc (ct))nf_ct_kill (ct);
@@ -795,6 +798,9 @@ __nf_conntrack_find_get(struct net *net, const struct nf_conntrack_zone *zone,*/ ct =nf_ct_tuplehash_to_ctrack (h);if (likely (refcount_inc_not_zero (&ct->ct_general.use))) { +/* re-check key after refcount */ +smp_acquire__after_ctrl_dep (); +if (likely (nf_ct_key_equal (h, tuple, zone, net))) goto found;
@@ -1387,6 +1393,9 @@ static unsigned int early_drop_list(struct net *net,if (!refcount_inc_not_zero (&tmp->ct_general.use))continue ; +/* load ->ct_net and ->status after refcount increase */ +smp_acquire__after_ctrl_dep (); +/* kill only if still in same netns -- might have moved due to * SLAB_TYPESAFE_BY_RCU rules. *
@@ -1536,6 +1545,9 @@ static void gc_worker(struct work_struct *work)if (!refcount_inc_not_zero (&tmp->ct_general.use))continue ; +/* load ->status after refcount increase */ +smp_acquire__after_ctrl_dep (); +if (gc_worker_skip_ct (tmp)) {nf_ct_put (tmp);continue ;
@@ -1775,6 +1787,16 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,if (!exp)__nf_ct_try_assign_helper (ct, tmpl,GFP_ATOMIC ); +/* Other CPU might have obtained a pointer to this object before it was +* released. Because refcount is 0, refcount_inc_not_zero() will fail. +* +* After refcount_set(1) it will succeed; ensure that zeroing of +* ct->status and the correct ct->net pointer are visible; else other +* core might observe CONFIRMED bit which means the entry is valid and +* in the hash table, but its not (anymore). +*/ +smp_wmb (); +/* Now it is going to be associated with an sk_buff, set refcount to 1. */ refcount_set (&ct->ct_general.use,1 );
net/netfilter/nf_conntrack_netlink.c +1 -0
@@ -1203,6 +1203,7 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb) hnnode) { ct =nf_ct_tuplehash_to_ctrack (h);if (nf_ct_is_expired (ct)) { +/* need to defer nf_ct_kill() until lock is released */ if (i <ARRAY_SIZE (nf_ct_evict) &&refcount_inc_not_zero (&ct->ct_general.use)) nf_ct_evict[i++] = ct;
net/netfilter/nf_conntrack_standalone.c +3 -0
@@ -306,6 +306,9 @@ static int ct_seq_show(struct seq_file *s, void *v)if (unlikely (!refcount_inc_not_zero (&ct->ct_general.use)))return 0 ; +/* load ->status after refcount increase */ +smp_acquire__after_ctrl_dep (); +if (nf_ct_should_gc (ct)) {nf_ct_kill (ct); goto release;
Signature · 1 line
2.30.2
Re: [PATCH net 1/3] netfilter: conntrack: fix crash due to confirmed bit load reordering
This series was applied to netdev/net.git (master) by Pablo Neira Ayuso <[email protected]>:
Hello:
This series was applied to netdev/net.git (master) by Pablo Neira Ayuso <[email protected]>:
Quoted 10 lines · On Mon, 11 Jul 2022 11:33:55 +0200 you wrote:
From: Florian Westphal <[email protected]>
Kajetan Puchalski reports crash on ARM, with backtrace of:
__nf_ct_delete_from_listsnf_ct_deleteearly_drop__nf_conntrack_alloc[...]
Here is the summary with links:
- [net,1/3] netfilter: conntrack: fix crash due to confirmed bit load reordering
https://git.kernel.org/netdev/net/c/0ed8f619b412 - [net,2/3] netfilter: nf_log: incorrect offset to network header https://git.kernel.org/netdev/net/c/7a847c00eeba - [net,3/3] netfilter: nf_tables: replace BUG_ON by element length check https://git.kernel.org/netdev/net/c/c39ba4de6b0a
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html
Pablo Neira Ayuso
maintainerPATCH 2/3 netfilter: nf_log: incorrect offset to network header
In reply to Pablo Neira AyusoNFPROTO_ARP is expecting to find the ARP header at the network offset.
In the particular case of ARP, HTYPE= field shows the initial bytes of the ethernet header destination MAC address.
netdev out: IN= OUT=bridge0 MACSRC=c2:76:e5:71:e1:de MACDST=36:b0:4a:e2:72:ea MACPROTO=0806 ARP HTYPE=14000 PTYPE=0x4ae2 OPCODE=49782
NFPROTO_NETDEV egress hook is also expecting to find the IP headers at the network offset.
--- net/netfilter/nf_log_syslog.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
net/netfilter/nf_log_syslog.c +4 -4
@@ -67,7 +67,7 @@ dump_arp_packet(struct nf_log_buf *m,unsigned int logflags;struct arphdr _arph; - ah =skb_header_pointer (skb,0 ,sizeof (_arph), &_arph); + ah =skb_header_pointer (skb, nhoff,sizeof (_arph), &_arph);if (!ah) {nf_log_buf_add (m,"TRUNCATED" );return ;
@@ -96,7 +96,7 @@ dump_arp_packet(struct nf_log_buf *m, ah->ar_pln != sizeof(__be32))return ; - ap =skb_header_pointer (skb,sizeof (_arph),sizeof (_arpp), &_arpp); + ap =skb_header_pointer (skb, nhoff +sizeof (_arph),sizeof (_arpp), &_arpp);if (!ap) {nf_log_buf_add (m," INCOMPLETE [%zu bytes]" , skb->len -sizeof (_arph));
@@ -149,7 +149,7 @@ static void nf_log_arp_packet(struct net *net, u_int8_t pf,nf_log_dump_packet_common (m, pf, hooknum, skb, in, out, loginfo, prefix); -dump_arp_packet (m, loginfo, skb,0 ); +dump_arp_packet (m, loginfo, skb,skb_network_offset (skb));nf_log_buf_close (m); }
@@ -850,7 +850,7 @@ static void nf_log_ip_packet(struct net *net, u_int8_t pf,if (in)dump_mac_header (m, loginfo, skb); -dump_ipv4_packet (net, m, loginfo, skb,0 ); +dump_ipv4_packet (net, m, loginfo, skb,skb_network_offset (skb));nf_log_buf_close (m); }
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 3/3 netfilter: nf_tables: replace BUG_ON by element length check
In reply to Pablo Neira AyusoBUG_ON can be triggered from userspace with an element with a large userdata area. Replace it by length check and return EINVAL instead. Over time extensions have been growing in size.
Pick a sufficiently old Fixes: tag to propagate this fix.
--- include/net/netfilter/nf_tables.h | 14 +++--- net/netfilter/nf_tables_api.c | 72 ++++++++++++++++++++++--------- 2 files changed, 60 insertions(+), 26 deletions(-)
include/net/netfilter/nf_tables.h +9 -5
@@ -657,18 +657,22 @@ static inline void nft_set_ext_prepare(struct nft_set_ext_tmpl *tmpl) tmpl->len =sizeof (struct nft_set_ext ); } -static inline void nft_set_ext_add_length (struct nft_set_ext_tmpl *tmpl,u8 id, -unsigned int len) +static inline int nft_set_ext_add_length (struct nft_set_ext_tmpl *tmpl,u8 id, +unsigned int len) { tmpl->len =ALIGN (tmpl->len, nft_set_ext_types[id].align); -BUG_ON (tmpl->len >U8_MAX ); +if (tmpl->len >U8_MAX ) +return -EINVAL ; + tmpl->offset[id] = tmpl->len; tmpl->len += nft_set_ext_types[id].len + len; + +return 0 ; } -static inline void nft_set_ext_add (struct nft_set_ext_tmpl *tmpl,u8 id) +static inline int nft_set_ext_add (struct nft_set_ext_tmpl *tmpl,u8 id) { -nft_set_ext_add_length (tmpl, id,0 ); +return nft_set_ext_add_length (tmpl, id,0 ); } static inline void nft_set_ext_init(struct nft_set_ext *ext,
net/netfilter/nf_tables_api.c +51 -21
148 lines not on this page · raw message
Signature · 1 line
2.30.2