linux-stable.git/include/net/netfilter, branch linux-4.9.y

netfilter: nf_queue: fix possible use-after-free

2022-03-08T18:00:57+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

netfilter: nat: remove l4 protocol port rovers

2022-02-08T17:15:28+00:00

commit 6ed5943f8735e2b778d92ea4d9805c0a1d89bc2b upstream.

This is a leftover from days where single-cpu systems were common:
Store last port used to resolve a clash to use it as a starting point when
the next conflict needs to be resolved.

When we have parallel attempt to connect to same address:port pair,
its likely that both cores end up computing the same "available" port,
as both use same starting port, and newly used ports won't become
visible to other cores until the conntrack gets confirmed later.

One of the cores then has to drop the packet at insertion time because
the chosen new tuple turns out to be in use after all.

Lets simplify this: remove port rover and use a pseudo-random starting
point.

Note that this doesn't make netfilter default to 'fully random' mode;
the 'rover' was only used if NAT could not reuse source port as-is.

Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Greg Kroah-Hartman

netfilter: nf_tables: fix destination register zeroing

2020-09-12T09:47:33+00:00

[ Upstream commit 1e105e6afa6c3d32bfb52c00ffa393894a525c27 ]

Following bug was reported via irc:
nft list ruleset
   set knock_candidates_ipv4 {
      type ipv4_addr . inet_service
      size 65535
      elements = { 127.0.0.1 . 123,
                   127.0.0.1 . 123 }
      }
 ..
   udp dport 123 add @knock_candidates_ipv4 { ip saddr . 123 }
   udp dport 123 add @knock_candidates_ipv4 { ip saddr . udp dport }

It should not have been possible to add a duplicate set entry.

After some debugging it turned out that the problem is the immediate
value (123) in the second-to-last rule.

Concatenations use 32bit registers, i.e. the elements are 8 bytes each,
not 6 and it turns out the kernel inserted

inet firewall @knock_candidates_ipv4
        element 0100007f ffff7b00  : 0 [end]
        element 0100007f 00007b00  : 0 [end]

Note the non-zero upper bits of the first element.  It turns out that
nft_immediate doesn't zero the destination register, but this is needed
when the length isn't a multiple of 4.

Furthermore, the zeroing in nft_payload is broken.  We can't use
[len / 4] = 0 -- if len is a multiple of 4, index is off by one.

Skip zeroing in this case and use a conditional instead of (len -1) / 4.

Fixes: 49499c3e6e18 ("netfilter: nf_tables: switch registers to 32 bit addressing")
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: conntrack: avoid gcc-10 zero-length-bounds warning

2020-05-20T06:15:36+00:00

[ Upstream commit 2c407aca64977ede9b9f35158e919773cae2082f ]

gcc-10 warns around a suspicious access to an empty struct member:

net/netfilter/nf_conntrack_core.c: In function '__nf_conntrack_alloc':
net/netfilter/nf_conntrack_core.c:1522:9: warning: array subscript 0 is outside the bounds of an interior zero-length array 'u8[0]' {aka 'unsigned char[0]'} [-Wzero-length-bounds]
 1522 |  memset(&ct->__nfct_init_offset[0], 0,
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from net/netfilter/nf_conntrack_core.c:37:
include/net/netfilter/nf_conntrack.h:90:5: note: while referencing '__nfct_init_offset'
   90 |  u8 __nfct_init_offset[0];
      |     ^~~~~~~~~~~~~~~~~~

The code is correct but a bit unusual. Rework it slightly in a way that
does not trigger the warning, using an empty struct instead of an empty
array. There are probably more elegant ways to do this, but this is the
smallest change.

Fixes: c41884ce0562 ("netfilter: conntrack: avoid zeroing timer")
Signed-off-by: Arnd Bergmann 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_tables: Align nft_expr private data to 64-bit

2019-11-12T18:15:45+00:00

commit 250367c59e6ba0d79d702a059712d66edacd4a1a upstream.

Invoking the following commands on a 32-bit architecture with strict
alignment requirements (such as an ARMv7-based Raspberry Pi) results
in an alignment exception:

 # nft add table ip test-ip4
 # nft add chain ip test-ip4 output { type filter hook output priority 0; }
 # nft add rule  ip test-ip4 output quota 1025 bytes

Alignment trap: not handling instruction e1b26f9f at [<7f4473f8>]
Unhandled fault: alignment exception (0x001) at 0xb832e824
Internal error: : 1 [#1] PREEMPT SMP ARM
Hardware name: BCM2835
[<7f4473fc>] (nft_quota_do_init [nft_quota])
[<7f447448>] (nft_quota_init [nft_quota])
[<7f4260d0>] (nf_tables_newrule [nf_tables])
[<7f4168dc>] (nfnetlink_rcv_batch [nfnetlink])
[<7f416bd0>] (nfnetlink_rcv [nfnetlink])
[<8078b334>] (netlink_unicast)
[<8078b664>] (netlink_sendmsg)
[<8071b47c>] (sock_sendmsg)
[<8071bd18>] (___sys_sendmsg)
[<8071ce3c>] (__sys_sendmsg)
[<8071ce94>] (sys_sendmsg)

The reason is that nft_quota_do_init() calls atomic64_set() on an
atomic64_t which is only aligned to 32-bit, not 64-bit, because it
succeeds struct nft_expr in memory which only contains a 32-bit pointer.
Fix by aligning the nft_expr private data to 64-bit.

Fixes: 96518518cc41 ("netfilter: add nftables")
Signed-off-by: Lukas Wunner 
Cc: stable@vger.kernel.org # v3.13+
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Greg Kroah-Hartman

netfilter: ctnetlink: don't use conntrack/expect object addresses as id

2019-08-25T08:51:42+00:00

commit 3c79107631db1f7fd32cf3f7368e4672004a3010 upstream.

else, we leak the addresses to userspace via ctnetlink events
and dumps.

Compute an ID on demand based on the immutable parts of nf_conn struct.

Another advantage compared to using an address is that there is no
immediate re-use of the same ID in case the conntrack entry is freed and
reallocated again immediately.

Fixes: 3583240249ef ("[NETFILTER]: nf_conntrack_expect: kill unique ID")
Fixes: 7f85f914721f ("[NETFILTER]: nf_conntrack: kill unique ID")
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Ben Hutchings 
Signed-off-by: Greg Kroah-Hartman

netfilter: physdev: relax br_netfilter dependency

2019-04-05T20:29:14+00:00

[ Upstream commit 8e2f311a68494a6677c1724bdcb10bada21af37c ]

Following command:
  iptables -D FORWARD -m physdev ...
causes connectivity loss in some setups.

Reason is that iptables userspace will probe kernel for the module revision
of the physdev patch, and physdev has an artificial dependency on
br_netfilter (xt_physdev use makes no sense unless a br_netfilter module
is loaded).

This causes the "phydev" module to be loaded, which in turn enables the
"call-iptables" infrastructure.

bridged packets might then get dropped by the iptables ruleset.

The better fix would be to change the "call-iptables" defaults to 0 and
enforce explicit setting to 1, but that breaks backwards compatibility.

This does the next best thing: add a request_module call to checkentry.
This was a stray '-D ... -m physdev' won't activate br_netfilter
anymore.

Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_tables: fix mismatch in big-endian system

2019-02-20T09:18:34+00:00

commit 10596608c4d62cb8c1c2b806debcbd32fe657e71 upstream.

Currently, there are two different methods to store an u16 integer to
the u32 data register. For example:
  u32 *dest = ®s->data[priv->dreg];
  1. *dest = 0; *(u16 *) dest = val_u16;
  2. *dest = val_u16;

For method 1, the u16 value will be stored like this, either in
big-endian or little-endian system:
  0          15           31
  +-+-+-+-+-+-+-+-+-+-+-+-+
  |   Value   |     0     |
  +-+-+-+-+-+-+-+-+-+-+-+-+

For method 2, in little-endian system, the u16 value will be the same
as listed above. But in big-endian system, the u16 value will be stored
like this:
  0          15           31
  +-+-+-+-+-+-+-+-+-+-+-+-+
  |     0     |   Value   |
  +-+-+-+-+-+-+-+-+-+-+-+-+

So later we use "memcmp(®s->data[priv->sreg], data, 2);" to do
compare in nft_cmp, nft_lookup expr ..., method 2 will get the wrong
result in big-endian system, as 0~15 bits will always be zero.

For the similar reason, when loading an u16 value from the u32 data
register, we should use "*(u16 *) sreg;" instead of "(u16)*sreg;",
the 2nd method will get the wrong value in the big-endian system.

So introduce some wrapper functions to store/load an u8 or u16
integer to/from the u32 data register, and use them in the right
place.

Signed-off-by: Liping Zhang 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

netfilter: nat: Revert "netfilter: nat: convert nat bysrc hash to rhashtable"

2017-11-18T10:22:24+00:00

commit e1bf1687740ce1a3598a1c5e452b852ff2190682 upstream.

This reverts commit 870190a9ec9075205c0fa795a09fa931694a3ff1.

It was not a good idea. The custom hash table was a much better
fit for this purpose.

A fast lookup is not essential, in fact for most cases there is no lookup
at all because original tuple is not taken and can be used as-is.
What needs to be fast is insertion and deletion.

rhlist removal however requires a rhlist walk.
We can have thousands of entries in such a list if source port/addresses
are reused for multiple flows, if this happens removal requests are so
expensive that deletions of a few thousand flows can take several
seconds(!).

The advantages that we got from rhashtable are:
1) table auto-sizing
2) multiple locks

1) would be nice to have, but it is not essential as we have at
most one lookup per new flow, so even a million flows in the bysource
table are not a problem compared to current deletion cost.
2) is easy to add to custom hash table.

I tried to add hlist_node to rhlist to speed up rhltable_remove but this
isn't doable without changing semantics.  rhltable_remove_fast will
check that the to-be-deleted object is part of the table and that
requires a list walk that we want to avoid.

Furthermore, using hlist_node increases size of struct rhlist_head, which
in turn increases nf_conn size.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=196821
Reported-by: Ivan Babrou 
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Greg Kroah-Hartman

netfilter: nf_tables: set pktinfo->thoff at AH header if found

2017-10-08T08:26:11+00:00

[ Upstream commit 568af6de058cb2b0c5b98d98ffcf37cdc6bc38a7 ]

Phil Sutter reports that IPv6 AH header matching is broken. From
userspace, nft generates bytecode that expects to find the AH header at
NFT_PAYLOAD_TRANSPORT_HEADER both for IPv4 and IPv6. However,
pktinfo->thoff is set to the inner header after the AH header in IPv6,
while in IPv4 pktinfo->thoff points to the AH header indeed. This
behaviour is inconsistent. This patch fixes this problem by updating
ipv6_find_hdr() to get the IP6_FH_F_AUTH flag so this function stops at
the AH header, so both IPv4 and IPv6 pktinfo->thoff point to the AH
header.

This is also inconsistent when trying to match encapsulated headers:

1) A packet that looks like IPv4 + AH + TCP dport 22 will *not* match.
2) A packet that looks like IPv6 + AH + TCP dport 22 will match.

Reported-by: Phil Sutter 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin 
Signed-off-by: Greg Kroah-Hartman