kernel_samsung_a53x/net
Stanislav Fomichev 868e067240 net: set SOCK_RCU_FREE before inserting socket into hashtable
commit 871019b22d1bcc9fab2d1feba1b9a564acbb6e99 upstream.

We've started to see the following kernel traces:

 WARNING: CPU: 83 PID: 0 at net/core/filter.c:6641 sk_lookup+0x1bd/0x1d0

 Call Trace:
  <IRQ>
  __bpf_skc_lookup+0x10d/0x120
  bpf_sk_lookup+0x48/0xd0
  bpf_sk_lookup_tcp+0x19/0x20
  bpf_prog_<redacted>+0x37c/0x16a3
  cls_bpf_classify+0x205/0x2e0
  tcf_classify+0x92/0x160
  __netif_receive_skb_core+0xe52/0xf10
  __netif_receive_skb_list_core+0x96/0x2b0
  napi_complete_done+0x7b5/0xb70
  <redacted>_poll+0x94/0xb0
  net_rx_action+0x163/0x1d70
  __do_softirq+0xdc/0x32e
  asm_call_irq_on_stack+0x12/0x20
  </IRQ>
  do_softirq_own_stack+0x36/0x50
  do_softirq+0x44/0x70

__inet_hash can race with lockless (rcu) readers on the other cpus:

  __inet_hash
    __sk_nulls_add_node_rcu
    <- (bpf triggers here)
    sock_set_flag(SOCK_RCU_FREE)

Let's move the SOCK_RCU_FREE part up a bit, before we are inserting
the socket into hashtables. Note, that the race is really harmless;
the bpf callers are handling this situation (where listener socket
doesn't have SOCK_RCU_FREE set) correctly, so the only
annoyance is a WARN_ONCE.

More details from Eric regarding SOCK_RCU_FREE timeline:

Commit 3b24d854cb35 ("tcp/dccp: do not touch listener sk_refcnt under
synflood") added SOCK_RCU_FREE. At that time, the precise location of
sock_set_flag(sk, SOCK_RCU_FREE) did not matter, because the thread calling
__inet_hash() owns a reference on sk. SOCK_RCU_FREE was only tested
at dismantle time.

Commit 6acc9b432e67 ("bpf: Add helper to retrieve socket in BPF")
started checking SOCK_RCU_FREE _after_ the lookup to infer whether
the refcount has been taken care of.

Fixes: 6acc9b432e67 ("bpf: Add helper to retrieve socket in BPF")
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[Resolved conflict for 5.10 and below.]
Signed-off-by: Siddh Raman Pant <siddh.raman.pant@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2025-01-19 14:52:20 +01:00
..
6lowpan
9p 9p/xen: fix release of IRQ 2024-12-17 13:24:22 +01:00
802
8021q gro: remove rcu_read_lock/rcu_read_unlock from gro_receive handlers 2025-01-19 00:10:01 +01:00
appletalk
atm
ax25 Revert "Make more sysctl constants read-only" 2024-12-03 19:56:17 +01:00
batman-adv batman-adv: Do not let TT changes list grows indefinitely 2025-01-02 17:00:49 +01:00
bluetooth Bluetooth: MGMT: Ignore keys being loaded with invalid type 2025-01-19 14:50:17 +01:00
bpf
bpfilter
bridge net: bridge: br_fdb_external_learn_add(): always set EXT_LEARN 2025-01-19 00:10:01 +01:00
caif
can can: bcm: Remove proc entry when dev is unregistered. 2025-01-19 00:10:01 +01:00
ceph libceph: fix race between delayed_work() and ceph_monc_stop() 2024-11-19 14:19:45 +01:00
core Revert "skb_expand_head() adjust skb->truesize incorrectly" 2025-01-15 16:38:33 +01:00
dcb
dccp dccp: Fix memory leak in dccp_feat_change_recv 2024-12-17 13:24:26 +01:00
decnet
dns_resolver
dsa
ethernet gro: remove rcu_read_lock/rcu_read_unlock from gro_receive handlers 2025-01-19 00:10:01 +01:00
ethtool ethtool: Fix wrong mod state in case of verbose and no_mask bitset 2024-12-17 13:24:27 +01:00
hsr net: hsr: avoid potential out-of-bound access in fill_frame_info() 2024-12-17 13:24:26 +01:00
ieee802154 net: ieee802154: do not leave a dangling sk pointer in ieee802154_create() 2024-12-17 13:24:30 +01:00
ife
ipv4 net: set SOCK_RCU_FREE before inserting socket into hashtable 2025-01-19 14:52:20 +01:00
ipv6 gro: remove rcu_read_lock/rcu_read_unlock from gro_receive handlers 2025-01-19 00:10:01 +01:00
iucv Revert "net/iucv: fix use after free in iucv_sock_close()" 2024-11-24 00:23:55 +01:00
kcm kcm: Serialise kcm_sendmsg() for the same socket. 2024-11-23 23:20:48 +01:00
key
l2tp genetlink: hold RCU in genlmsg_mcast() 2024-11-23 23:21:59 +01:00
l3mdev net: Add l3mdev index to flow struct and avoid oif reset for port devices 2024-11-23 23:21:52 +01:00
lapb
llc net: llc: reset skb->transport_header 2025-01-15 16:29:53 +01:00
mac80211 wifi: mac80211: wake the queues in case of failure in resume 2025-01-15 16:29:54 +01:00
mac802154 Revert "net: mac802154: Fix racy device stats updates by DEV_STATS_INC() and DEV_STATS_ADD()" 2024-11-19 14:52:14 +01:00
mpls
mptcp mptcp: pr_debug: add missing \n at the end 2025-01-19 14:48:42 +01:00
ncm
ncsi net/ncsi: Fix the multi thread manner of NCSI driver 2024-11-19 14:19:00 +01:00
netfilter netfilter: ipset: Fix for recursive locking warning 2025-01-15 16:29:41 +01:00
netlabel
netlink netlink: terminate outstanding dump on socket close 2024-12-17 13:20:50 +01:00
netrom netrom: check buffer length before accessing it 2025-01-15 16:29:53 +01:00
nfc
nsh
openvswitch
packet af_packet: fix vlan_get_protocol_dgram() vs MSG_PEEK 2025-01-15 16:29:54 +01:00
phonet Revert "Make more sysctl constants read-only" 2024-12-03 19:56:17 +01:00
psample
qrtr Revert "net: qrtr: Update packets cloning when broadcasting" 2024-11-24 00:23:18 +01:00
rds Revert "net:rds: Fix possible deadlock in rds_message_put" 2024-11-24 00:23:49 +01:00
rfkill net: rfkill: gpio: Add check for clk_enable() 2024-12-17 13:24:07 +01:00
rose Revert "Make more sysctl constants read-only" 2024-12-03 19:56:17 +01:00
rxrpc
sched sch/netem: fix use after free in netem_dequeue 2025-01-19 14:51:17 +01:00
sctp net/sctp: Prevent autoclose integer overflow in sctp_association_init() 2025-01-15 16:29:56 +01:00
skb_tracer
smc net/smc: check return value of sock_recvmsg when draining clc data 2025-01-15 16:29:41 +01:00
strparser
sunrpc sunrpc: don't change ->sv_stats if it doesn't exist 2025-01-19 14:48:42 +01:00
switchdev
tipc tipc: fix NULL deref in cleanup_bearer() 2025-01-02 17:00:49 +01:00
tls
unix Revert "af_unix: Remove put_pid()/put_cred() in copy_peercred()." 2024-11-24 00:23:43 +01:00
vmw_vsock virtio/vsock: Fix accept_queue memory leak 2025-01-02 17:00:49 +01:00
wimax
wireless Revert "wifi: nl80211: don't give key data to userspace" 2024-11-24 00:23:55 +01:00
x25 Revert "Make more sysctl constants read-only" 2024-12-03 19:56:17 +01:00
xdp
xfrm xfrm: store and rely on direction to construct offload flags 2024-12-17 13:24:04 +01:00
compat.c
devres.c
Kconfig
Makefile
socket.c
sysctl_net.c
TEST_MAPPING