The selftests test_maps program was leaving dangling BPF sockmap programs around because not all psock elements were removed from the map. The elements in turn hold a reference on the BPF program they are attached to causing BPF programs to stay open even after test_maps has completed. The original intent was that sk_state_change() would be called when TCP socks went through TCP_CLOSE state. However, because socks may be in SOCK_DEAD state or the sock may be a listening socket the event is not always triggered. To resolve this use the ULP infrastructure and register our own proto close() handler. This fixes the above case.
Showing the changes to include/net (1 of 2 files). Show all
Unified · Split
| @@ -1985,6 +1985,7 @@ enum hrtimer_restart tcp_pace_kick(struct hrtimer *timer); · 1984 unmodified lines | |||
| 1985 | 1985 | ||
| 1986 | enum { | 1986 | enum { |
| 1987 | TCP_ULP_TLS, | 1987 | TCP_ULP_TLS, |
| 1988 | TCP_ULP_BPF, | ||
| 1988 | }; | 1989 | }; |
| 1989 | 1990 | ||
| 1990 | struct tcp_ulp_ops { | 1991 | struct tcp_ulp_ops { |
| @@ -2003,6 +2004,7 @@ struct tcp_ulp_ops { · 12 unmodified lines | |||
| 2003 | int tcp_register_ulp(struct tcp_ulp_ops *type); | 2004 | int tcp_register_ulp(struct tcp_ulp_ops *type); |
| 2004 | void tcp_unregister_ulp(struct tcp_ulp_ops *type); | 2005 | void tcp_unregister_ulp(struct tcp_ulp_ops *type); |
| 2005 | int tcp_set_ulp(struct sock *sk, const char *name); | 2006 | int tcp_set_ulp(struct sock *sk, const char *name); |
| 2007 | int tcp_set_ulp_id(struct sock *sk, const int ulp); | ||
| 2006 | void tcp_get_available_ulp(char *buf, size_t len); | 2008 | void tcp_get_available_ulp(char *buf, size_t len); |
| 2007 | void tcp_cleanup_ulp(struct sock *sk); | 2009 | void tcp_cleanup_ulp(struct sock *sk); |
| 2008 | 2010 | ||