ipvs: convert connection locking

Convert __ip_vs_conntbl_lock_array as follows:

- readers that do not modify conn lists will use RCU lock
- updaters that modify lists will use spinlock_t

Now for conn lookups we will use RCU read-side
critical section. Without using __ip_vs_conn_get such
places have access to connection fields and can
dereference some pointers like pe and pe_data plus
the ability to update timer expiration. If full access
is required we contend for reference.

We add barrier in __ip_vs_conn_put, so that
other CPUs see the refcnt operation after other writes.

With the introduction of ip_vs_conn_unlink()
we try to reorganize ip_vs_conn_expire(), so that
unhashing of connections that should stay more time is
avoided, even if it is for very short time.

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off by: Hans Schillstrom <hans@schillstrom.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
This commit is contained in:
Julian Anastasov
2013-03-21 11:58:10 +02:00
committed by Pablo Neira Ayuso
parent 60b6aa3b31
commit 088339a57d
2 changed files with 134 additions and 108 deletions

View File

@@ -620,6 +620,8 @@ struct ip_vs_conn {
const struct ip_vs_pe *pe;
char *pe_data;
__u8 pe_data_len;
struct rcu_head rcu_head;
};
/*
@@ -1185,9 +1187,19 @@ struct ip_vs_conn * ip_vs_conn_out_get_proto(int af, const struct sk_buff *skb,
const struct ip_vs_iphdr *iph,
int inverse);
/* Get reference to gain full access to conn.
* By default, RCU read-side critical sections have access only to
* conn fields and its PE data, see ip_vs_conn_rcu_free() for reference.
*/
static inline bool __ip_vs_conn_get(struct ip_vs_conn *cp)
{
return atomic_inc_not_zero(&cp->refcnt);
}
/* put back the conn without restarting its timer */
static inline void __ip_vs_conn_put(struct ip_vs_conn *cp)
{
smp_mb__before_atomic_dec();
atomic_dec(&cp->refcnt);
}
extern void ip_vs_conn_put(struct ip_vs_conn *cp);