tcp: fix possible deadlock in TCP stack vs BPF filter
authorEric Dumazet <edumazet@google.com>
Tue, 15 Aug 2017 00:44:43 +0000 (17:44 -0700)
committerDavid S. Miller <davem@davemloft.net>
Tue, 15 Aug 2017 05:31:27 +0000 (22:31 -0700)
commitd624d276d1ddacbcb12ad96832ce0c7b82cd25db
treeb78aa1f2060bf2aa70b13af5d58a7bca994af306
parent7749d4ff88d31b0be17c8683143135adaaadc6a7
tcp: fix possible deadlock in TCP stack vs BPF filter

Filtering the ACK packet was not put at the right place.

At this place, we already allocated a child and put it
into accept queue.

We absolutely need to call tcp_child_process() to release
its spinlock, or we will deadlock at accept() or close() time.

Found by syzkaller team (Thanks a lot !)

Fixes: 8fac365f63c8 ("tcp: Add a tcp_filter hook before handle ack packet")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Chenbo Feng <fengc@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/tcp_ipv4.c
net/ipv6/tcp_ipv6.c