tcp/dccp: fix possible race __inet_lookup_established()
authorEric Dumazet <edumazet@google.com>
Sat, 14 Dec 2019 02:20:41 +0000 (18:20 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 4 Jan 2020 13:00:19 +0000 (14:00 +0100)
commit94671cf125f8e08e093290451d511c197b005c82
treeb70d9691d0145f6d612f717750fe20b25d1fd314
parent0ddfaacfb76d8d54a1f94bfa4e4bc9a7c452d4c7
tcp/dccp: fix possible race __inet_lookup_established()

commit 8dbd76e79a16b45b2ccb01d2f2e08dbf64e71e40 upstream.

Michal Kubecek and Firo Yang did a very nice analysis of crashes
happening in __inet_lookup_established().

Since a TCP socket can go from TCP_ESTABLISH to TCP_LISTEN
(via a close()/socket()/listen() cycle) without a RCU grace period,
I should not have changed listeners linkage in their hash table.

They must use the nulls protocol (Documentation/RCU/rculist_nulls.txt),
so that a lookup can detect a socket in a hash list was moved in
another one.

Since we added code in commit d296ba60d8e2 ("soreuseport: Resolve
merge conflict for v4/v6 ordering fix"), we have to add
hlist_nulls_add_tail_rcu() helper.

Fixes: 3b24d854cb35 ("tcp/dccp: do not touch listener sk_refcnt under synflood")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Michal Kubecek <mkubecek@suse.cz>
Reported-by: Firo Yang <firo.yang@suse.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Link: https://lore.kernel.org/netdev/20191120083919.GH27852@unicorn.suse.cz/
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
[stable-4.14: we also need to update code in __inet_lookup_listener() and
 inet6_lookup_listener() which has been removed in 5.0-rc1.]
Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
include/linux/rculist_nulls.h
include/net/inet_hashtables.h
include/net/sock.h
net/ipv4/inet_diag.c
net/ipv4/inet_hashtables.c
net/ipv4/tcp_ipv4.c
net/ipv6/inet6_hashtables.c