linux-stable.git/net/netfilter, branch v5.3.9

netfilter: conntrack: avoid possible false sharing

2019-11-06T12:09:28+00:00

commit e37542ba111f3974dc622ae0a21c1787318de500 upstream.

As hinted by KCSAN, we need at least one READ_ONCE()
to prevent a compiler optimization.

More details on :
https://github.com/google/ktsan/wiki/READ_ONCE-and-WRITE_ONCE#it-may-improve-performance

sysbot report :
BUG: KCSAN: data-race in __nf_ct_refresh_acct / __nf_ct_refresh_acct

read to 0xffff888123eb4f08 of 4 bytes by interrupt on cpu 0:
 __nf_ct_refresh_acct+0xd4/0x1b0 net/netfilter/nf_conntrack_core.c:1796
 nf_ct_refresh_acct include/net/netfilter/nf_conntrack.h:201 [inline]
 nf_conntrack_tcp_packet+0xd40/0x3390 net/netfilter/nf_conntrack_proto_tcp.c:1161
 nf_conntrack_handle_packet net/netfilter/nf_conntrack_core.c:1633 [inline]
 nf_conntrack_in+0x410/0xaa0 net/netfilter/nf_conntrack_core.c:1727
 ipv4_conntrack_in+0x27/0x40 net/netfilter/nf_conntrack_proto.c:178
 nf_hook_entry_hookfn include/linux/netfilter.h:135 [inline]
 nf_hook_slow+0x83/0x160 net/netfilter/core.c:512
 nf_hook include/linux/netfilter.h:260 [inline]
 NF_HOOK include/linux/netfilter.h:303 [inline]
 ip_rcv+0x12f/0x1a0 net/ipv4/ip_input.c:523
 __netif_receive_skb_one_core+0xa7/0xe0 net/core/dev.c:5004
 __netif_receive_skb+0x37/0xf0 net/core/dev.c:5118
 netif_receive_skb_internal+0x59/0x190 net/core/dev.c:5208
 napi_skb_finish net/core/dev.c:5671 [inline]
 napi_gro_receive+0x28f/0x330 net/core/dev.c:5704
 receive_buf+0x284/0x30b0 drivers/net/virtio_net.c:1061
 virtnet_receive drivers/net/virtio_net.c:1323 [inline]
 virtnet_poll+0x436/0x7d0 drivers/net/virtio_net.c:1428
 napi_poll net/core/dev.c:6352 [inline]
 net_rx_action+0x3ae/0xa50 net/core/dev.c:6418
 __do_softirq+0x115/0x33f kernel/softirq.c:292

write to 0xffff888123eb4f08 of 4 bytes by task 7191 on cpu 1:
 __nf_ct_refresh_acct+0xfb/0x1b0 net/netfilter/nf_conntrack_core.c:1797
 nf_ct_refresh_acct include/net/netfilter/nf_conntrack.h:201 [inline]
 nf_conntrack_tcp_packet+0xd40/0x3390 net/netfilter/nf_conntrack_proto_tcp.c:1161
 nf_conntrack_handle_packet net/netfilter/nf_conntrack_core.c:1633 [inline]
 nf_conntrack_in+0x410/0xaa0 net/netfilter/nf_conntrack_core.c:1727
 ipv4_conntrack_local+0xbe/0x130 net/netfilter/nf_conntrack_proto.c:200
 nf_hook_entry_hookfn include/linux/netfilter.h:135 [inline]
 nf_hook_slow+0x83/0x160 net/netfilter/core.c:512
 nf_hook include/linux/netfilter.h:260 [inline]
 __ip_local_out+0x1f7/0x2b0 net/ipv4/ip_output.c:114
 ip_local_out+0x31/0x90 net/ipv4/ip_output.c:123
 __ip_queue_xmit+0x3a8/0xa40 net/ipv4/ip_output.c:532
 ip_queue_xmit+0x45/0x60 include/net/ip.h:236
 __tcp_transmit_skb+0xdeb/0x1cd0 net/ipv4/tcp_output.c:1158
 __tcp_send_ack+0x246/0x300 net/ipv4/tcp_output.c:3685
 tcp_send_ack+0x34/0x40 net/ipv4/tcp_output.c:3691
 tcp_cleanup_rbuf+0x130/0x360 net/ipv4/tcp.c:1575

Reported by Kernel Concurrency Sanitizer on:
CPU: 1 PID: 7191 Comm: syz-fuzzer Not tainted 5.3.0+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011

Fixes: cc16921351d8 ("netfilter: conntrack: avoid same-timeout update")
Signed-off-by: Eric Dumazet 
Reported-by: syzbot 
Cc: Jozsef Kadlecsik 
Cc: Florian Westphal 
Acked-by: Pablo Neira Ayuso 
Signed-off-by: Jakub Kicinski 
Signed-off-by: Greg Kroah-Hartman

netfilter: nft_connlimit: disable bh on garbage collection

2019-10-29T08:21:36+00:00

[ Upstream commit 34a4c95abd25ab41fb390b985a08a651b1fa0b0f ]

BH must be disabled when invoking nf_conncount_gc_list() to perform
garbage collection, otherwise deadlock might happen.

  nf_conncount_add+0x1f/0x50 [nf_conncount]
  nft_connlimit_eval+0x4c/0xe0 [nft_connlimit]
  nft_dynset_eval+0xb5/0x100 [nf_tables]
  nft_do_chain+0xea/0x420 [nf_tables]
  ? sch_direct_xmit+0x111/0x360
  ? noqueue_init+0x10/0x10
  ? __qdisc_run+0x84/0x510
  ? tcp_packet+0x655/0x1610 [nf_conntrack]
  ? ip_finish_output2+0x1a7/0x430
  ? tcp_error+0x130/0x150 [nf_conntrack]
  ? nf_conntrack_in+0x1fc/0x4c0 [nf_conntrack]
  nft_do_chain_ipv4+0x66/0x80 [nf_tables]
  nf_hook_slow+0x44/0xc0
  ip_rcv+0xb5/0xd0
  ? ip_rcv_finish_core.isra.19+0x360/0x360
  __netif_receive_skb_one_core+0x52/0x70
  netif_receive_skb_internal+0x34/0xe0
  napi_gro_receive+0xba/0xe0
  e1000_clean_rx_irq+0x1e9/0x420 [e1000e]
  e1000e_poll+0xbe/0x290 [e1000e]
  net_rx_action+0x149/0x3b0
  __do_softirq+0xde/0x2d8
  irq_exit+0xba/0xc0
  do_IRQ+0x85/0xd0
  common_interrupt+0xf/0xf
  
  RIP: 0010:nf_conncount_gc_list+0x3b/0x130 [nf_conncount]

Fixes: 2f971a8f4255 ("netfilter: nf_conncount: move all list iterations under spinlock")
Reported-by: Laura Garcia Liebana 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_tables: allow lookups in dynamic sets

2019-10-11T16:36:34+00:00

[ Upstream commit acab713177377d9e0889c46bac7ff0cfb9a90c4d ]

This un-breaks lookups in sets that have the 'dynamic' flag set.
Given this active example configuration:

table filter {
  set set1 {
    type ipv4_addr
    size 64
    flags dynamic,timeout
    timeout 1m
  }

  chain input {
     type filter hook input priority 0; policy accept;
  }
}

... this works:
nft add rule ip filter input add @set1 { ip saddr }

-> whenever rule is triggered, the source ip address is inserted
into the set (if it did not exist).

This won't work:
nft add rule ip filter input ip saddr @set1 counter
Error: Could not process rule: Operation not supported

In other words, we can add entries to the set, but then can't make
matching decision based on that set.

That is just wrong -- all set backends support lookups (else they would
not be very useful).
The failure comes from an explicit rejection in nft_lookup.c.

Looking at the history, it seems like NFT_SET_EVAL used to mean
'set contains expressions' (aka. "is a meter"), for instance something like

 nft add rule ip filter input meter example { ip saddr limit rate 10/second }
 or
 nft add rule ip filter input meter example { ip saddr counter }

The actual meaning of NFT_SET_EVAL however, is
'set can be updated from the packet path'.

'meters' and packet-path insertions into sets, such as
'add @set { ip saddr }' use exactly the same kernel code (nft_dynset.c)
and thus require a set backend that provides the ->update() function.

The only set that provides this also is the only one that has the
NFT_SET_EVAL feature flag.

Removing the wrong check makes the above example work.
While at it, also fix the flag check during set instantiation to
allow supported combinations only.

Fixes: 8aeff920dcc9b3f ("netfilter: nf_tables: add stateful object reference to set elements")
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso 
Signed-off-by: Sasha Levin

netfilter: nf_flow_table: set default timeout after successful insertion

2019-09-03T20:55:42+00:00

Set up the default timeout for this new entry otherwise the garbage
collector might quickly remove it right after the flowtable insertion.

Fixes: ac2a66665e23 ("netfilter: add generic flow table infrastructure")
Signed-off-by: Pablo Neira Ayuso

netfilter: ctnetlink: honor IPS_OFFLOAD flag

2019-09-03T20:55:41+00:00

If this flag is set, timeout and state are irrelevant to userspace.

Fixes: 90964016e5d3 ("netfilter: nf_conntrack: add IPS_OFFLOAD status bit")
Signed-off-by: Pablo Neira Ayuso

netfilter: nft_fib_netdev: Terminate rule eval if protocol=IPv6 and ipv6 module is disabled

2019-09-03T20:53:56+00:00

If IPv6 is disabled on boot (ipv6.disable=1), but nft_fib_inet ends up
dealing with a IPv6 packet, it causes a kernel panic in
fib6_node_lookup_1(), crashing in bad_page_fault.

The panic is caused by trying to deference a very low address (0x38
in ppc64le), due to ipv6.fib6_main_tbl = NULL.
BUG: Kernel NULL pointer dereference at 0x00000038

The kernel panic was reproduced in a host that disabled IPv6 on boot and
have to process guest packets (coming from a bridge) using it's ip6tables.

Terminate rule evaluation when packet protocol is IPv6 but the ipv6 module
is not loaded.

Signed-off-by: Leonardo Bras 
Acked-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso

netfilter: nft_socket: fix erroneous socket assignment

2019-09-02T21:20:59+00:00

The socket assignment is wrong, see skb_orphan():
When skb->destructor callback is not set, but skb->sk is set, this hits BUG().

Link: https://bugzilla.redhat.com/show_bug.cgi?id=1651813
Fixes: 554ced0a6e29 ("netfilter: nf_tables: add support for native socket matching")
Signed-off-by: Fernando Fernandez Mancera 
Signed-off-by: Pablo Neira Ayuso

netfilter: nf_flow_table: clear skb tstamp before xmit

2019-08-29T14:38:05+00:00

If 'fq' qdisc is used and a program has requested timestamps,
skb->tstamp needs to be cleared, else fq will treat these as
'transmit time'.

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

netfilter: conntrack: make sysctls per-namespace again

2019-08-27T15:46:13+00:00

When I merged the extension sysctl tables with the main one I forgot to
reset them on netns creation.  They currently read/write init_net settings.

Fixes: d912dec12428 ("netfilter: conntrack: merge acct and helper sysctl table with main one")
Fixes: cb2833ed0044 ("netfilter: conntrack: merge ecache and timestamp sysctl tables with main one")
Reported-by: Shmulik Ladkani 
Signed-off-by: Florian Westphal 
Signed-off-by: Pablo Neira Ayuso

netfilter: nf_conntrack_ftp: Fix debug output

2019-08-27T11:01:05+00:00

The find_pattern() debug output was printing the 'skip' character.
This can be a NULL-byte and messes up further pr_debug() output.

Output without the fix:
kernel: nf_conntrack_ftp: Pattern matches!
kernel: nf_conntrack_ftp: Skipped up to `<7>nf_conntrack_ftp: find_pattern `PORT': dlen = 8
kernel: nf_conntrack_ftp: find_pattern `EPRT': dlen = 8

Output with the fix:
kernel: nf_conntrack_ftp: Pattern matches!
kernel: nf_conntrack_ftp: Skipped up to 0x0 delimiter!
kernel: nf_conntrack_ftp: Match succeeded!
kernel: nf_conntrack_ftp: conntrack_ftp: match `172,17,0,100,200,207' (20 bytes at 4150681645)
kernel: nf_conntrack_ftp: find_pattern `PORT': dlen = 8

Signed-off-by: Thomas Jarosch 
Signed-off-by: Pablo Neira Ayuso