linux-stable.git/net/netfilter, branch linux-5.16.y

netfilter: conntrack: revisit gc autotuning

2022-04-13T18:03:07+00:00

[ Upstream commit 2cfadb761d3d0219412fd8150faea60c7e863833 ]

as of commit 4608fdfc07e1
("netfilter: conntrack: collect all entries in one cycle")
conntrack gc was changed to run every 2 minutes.

On systems where conntrack hash table is set to large value, most evictions
happen from gc worker rather than the packet path due to hash table
distribution.

This causes netlink event overflows when events are collected.

This change collects average expiry of scanned entries and
reschedules to the average remaining value, within 1 to 60 second interval.

To avoid event overflows, reschedule after each bucket and add a
limit for both run time and number of evictions per run.

If more entries have to be evicted, reschedule and restart 1 jiffy
into the future.

Reported-by: Karel Rericha 
Cc: Shmulik Ladkani 
Cc: Eyal Birger 
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

net: prefer nf_ct_put instead of nf_conntrack_put

2022-04-08T12:06:31+00:00

[ Upstream commit 408bdcfce8dfd6902f75fbcd3b99d8b24b506597 ]

Its the same as nf_conntrack_put(), but without the
need for an indirect call.  The downside is a module dependency on
nf_conntrack, but all of these already depend on conntrack anyway.

Cc: Paul Blakey 
Cc: dev@openvswitch.org
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_conntrack_tcp: preserve liberal flag in tcp options

2022-04-08T12:06:28+00:00

[ Upstream commit f2dd495a8d589371289981d5ed33e6873df94ecc ]

Do not reset IP_CT_TCP_FLAG_BE_LIBERAL flag in out-of-sync scenarios
coming before the TCP window tracking, otherwise such connections will
fail in the window check.

Update tcp_options() to leave this flag in place and add a new helper
function to reset the tcp window state.

Based on patch from Sven Auhagen.

Fixes: c4832c7bbc3f ("netfilter: nf_ct_tcp: improve out-of-sync situation in TCP tracking")
Tested-by: Sven Auhagen 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: conntrack: Add and use nf_ct_set_auto_assign_helper_warned()

2022-04-08T12:06:19+00:00

[ Upstream commit 31d0bb9763efad30377505f3467f958d1ebe1e3d ]

The function sets the pernet boolean to avoid the spurious warning from
nf_ct_lookup_helper() when assigning conntrack helpers via nftables.

Fixes: 1a64edf54f55 ("netfilter: nft_ct: add helper set support")
Signed-off-by: Phil Sutter 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: flowtable: Fix QinQ and pppoe support for inet table

2022-04-08T12:06:17+00:00

[ Upstream commit 0492d857636e1c52cd71594a723c4b26a7b31978 ]

nf_flow_offload_inet_hook() does not check for 802.1q and PPPoE.
Fetch inner ethertype from these encapsulation protocols.

Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
Fixes: 4cd91f7c290f ("netfilter: flowtable: add vlan support")
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_tables: validate registers coming from userspace.

2022-03-28T07:59:55+00:00

commit 6e1acfa387b9ff82cfc7db8cc3b6959221a95851 upstream.

Bail out in case userspace uses unsupported registers.

Fixes: 49499c3e6e18 ("netfilter: nf_tables: switch registers to 32 bit addressing")
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Greg Kroah-Hartman

netfilter: nf_tables: initialize registers in nft_do_chain()

2022-03-28T07:59:54+00:00

commit 4c905f6740a365464e91467aa50916555b28213d upstream.

Initialize registers to avoid stack leak into userspace.

Fixes: 96518518cc41 ("netfilter: add nftables")
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Greg Kroah-Hartman

netfilter: nf_tables: prefer kfree_rcu(ptr, rcu) variant

2022-03-08T18:14:14+00:00

[ Upstream commit ae089831ff28a115908b8d796f667c2dadef1637 ]

While kfree_rcu(ptr) _is_ supported, it has some limitations.

Given that 99.99% of kfree_rcu() users [1] use the legacy
two parameters variant, and @catchall objects do have an rcu head,
simply use it.

Choice of kfree_rcu(ptr) variant was probably not intentional.

[1] including calls from net/netfilter/nf_tables_api.c

Fixes: aaa31047a6d2 ("netfilter: nftables: add catch-all set element support")
Signed-off-by: Eric Dumazet 
Reviewed-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_queue: handle socket prefetch

2022-03-08T18:14:08+00:00

commit 3b836da4081fa585cf6c392f62557496f2cb0efe upstream.

In case someone combines bpf socket assign and nf_queue, then we will
queue an skb who references a struct sock that did not have its
reference count incremented.

As we leave rcu protection, there is no guarantee that skb->sk is still
valid.

For refcount-less skb->sk case, try to increment the reference count
and then override the destructor.

In case of failure we have two choices: orphan the skb and 'delete'
preselect or let nf_queue() drop the packet.

Do the latter, it should not happen during normal operation.

Fixes: cf7fbe660f2d ("bpf: Add socket assign support")
Acked-by: Joe Stringer 
Signed-off-by: Florian Westphal 
Signed-off-by: Greg Kroah-Hartman

netfilter: nf_queue: fix possible use-after-free

2022-03-08T18:14:08+00:00

commit c3873070247d9e3c7a6b0cf9bf9b45e8018427b1 upstream.

Eric Dumazet says:
  The sock_hold() side seems suspect, because there is no guarantee
  that sk_refcnt is not already 0.

On failure, we cannot queue the packet and need to indicate an
error.  The packet will be dropped by the caller.

v2: split skb prefetch hunk into separate change

Fixes: 271b72c7fa82c ("udp: RCU handling for Unicast packets.")
Reported-by: Eric Dumazet 
Reviewed-by: Eric Dumazet 
Signed-off-by: Florian Westphal 
Signed-off-by: Greg Kroah-Hartman