Hello, my apologize to resend the same diff [1]. I'm not sure I got OK or not. It can be the case the privately received OK got lost.
The change is required to allow multiple instances of pf_test() running concurrently. Without this change in, PF trips 'KASSERT(sk->reverse == NULL);' thanks and regards sashan [1] http://openbsd-archive.7691.n7.nabble.com/pf-state-key-link-reverse-needs-atomic-ops-td367859.html --------8<---------------8<---------------8<------------------8<-------- diff --git a/sys/net/pf.c b/sys/net/pf.c index ebe339921fa..738759e556c 100644 --- a/sys/net/pf.c +++ b/sys/net/pf.c @@ -7398,11 +7398,20 @@ pf_inp_unlink(struct inpcb *inp) void pf_state_key_link_reverse(struct pf_state_key *sk, struct pf_state_key *skrev) { - /* Note that sk and skrev may be equal, then we refcount twice. */ - KASSERT(sk->reverse == NULL); - KASSERT(skrev->reverse == NULL); - sk->reverse = pf_state_key_ref(skrev); - skrev->reverse = pf_state_key_ref(sk); + struct pf_state_key *old_reverse; + + old_reverse = atomic_cas_ptr(&sk->reverse, NULL, skrev); + if (old_reverse != NULL) + KASSERT(old_reverse == skrev); + else + pf_state_key_ref(skrev); + + + old_reverse = atomic_cas_ptr(&skrev->reverse, NULL, sk); + if (old_reverse != NULL) + KASSERT(old_reverse == sk); + else + pf_state_key_ref(sk); } #if NPFLOG > 0
