ipvs: use resizable hash table for services

Make the hash table for services resizable in the bit range of 4-20.
Table is attached only while services are present. Resizing is done
by delayed work based on load (the number of hashed services).
Table grows when load increases 2+ times (above 12.5% with lfactor=-3)
and shrinks 8+ times when load decreases 16+ times (below 0.78%).

Switch to jhash hashing to reduce the collisions for multiple
services.

Add a hash_key field into the service to store the table ID in
the highest bit and the entry's hash value in the lowest bits. The
lowest part of the hash value is used as bucket ID, the remaining
part is used to filter the entries in the bucket before matching
the keys and as result, helps the lookup operation to access only
one cache line. By knowing the table ID and bucket ID for entry,
we can unlink it without calculating the hash value and doing
lookup by keys. We need only to validate the saved hash_key under
lock.

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Florian Westphal <fw@strlen.de>

1 files changed, 34 insertions, 15 deletions

diff --git a/include/net/ip_vs.h b/include/net/ip_vs.h
index c373fbdd2d0f..663ad6ad9518 100644
--- a/include/net/ip_vs.h
+++ b/include/net/ip_vs.h
@@ -35,12 +35,10 @@
 
 #define IP_VS_HDR_INVERSE	1
 #define IP_VS_HDR_ICMP		2
-/*
- *	Hash table: for virtual service lookups
- */
-#define IP_VS_SVC_TAB_BITS 8
-#define IP_VS_SVC_TAB_SIZE BIT(IP_VS_SVC_TAB_BITS)
-#define IP_VS_SVC_TAB_MASK (IP_VS_SVC_TAB_SIZE - 1)
+
+/* svc_table limits */
+#define IP_VS_SVC_TAB_MIN_BITS	4
+#define IP_VS_SVC_TAB_MAX_BITS	20
 
 /* Generic access of ipvs struct */
 static inline struct netns_ipvs *net_ipvs(struct net* net)
@@ -51,8 +49,6 @@ static inline struct netns_ipvs *net_ipvs(struct net* net)
 /* Connections' size value needed by ip_vs_ctl.c */
 extern int ip_vs_conn_tab_size;
 
-extern struct mutex __ip_vs_mutex;
-
 struct ip_vs_iphdr {
 	int hdr_flags;	/* ipvs flags */
 	__u32 off;	/* Where IP or IPv4 header starts */
@@ -289,6 +285,12 @@ static inline int ip_vs_af_index(int af)
 	return af == AF_INET6 ? IP_VS_AF_INET6 : IP_VS_AF_INET;
 }
 
+/* work_flags */
+enum {
+	IP_VS_WORK_SVC_RESIZE,		/* Schedule svc_resize_work */
+	IP_VS_WORK_SVC_NORESIZE,	/* Stopping svc_resize_work */
+};
+
 /* The port number of FTP service (in network order). */
 #define FTPPORT  cpu_to_be16(21)
 #define FTPDATA  cpu_to_be16(20)
@@ -889,14 +891,15 @@ struct ip_vs_dest_user_kern {
  * forwarding entries.
  */
 struct ip_vs_service {
-	struct hlist_node	s_list;   /* node in service table */
-	atomic_t		refcnt;   /* reference counter */
-
+	struct hlist_bl_node	s_list;   /* node in service table */
+	u32			hash_key; /* Key for the hash table */
 	u16			af;       /* address family */
 	__u16			protocol; /* which protocol (TCP/UDP) */
+
 	union nf_inet_addr	addr;	  /* IP address for virtual service */
-	__be16			port;	  /* port number for the service */
 	__u32                   fwmark;   /* firewall mark of the service */
+	atomic_t		refcnt;   /* reference counter */
+	__be16			port;	  /* port number for the service */
 	unsigned int		flags;	  /* service status flags */
 	unsigned int		timeout;  /* persistent timeout in ticks */
 	__be32			netmask;  /* grouping granularity, mask/plen */
@@ -1155,6 +1158,10 @@ struct netns_ipvs {
 	struct list_head	dest_trash;
 	spinlock_t		dest_trash_lock;
 	struct timer_list	dest_trash_timer; /* expiration timer */
+	struct mutex		service_mutex;    /* service reconfig */
+	struct rw_semaphore	svc_resize_sem;   /* svc_table resizing */
+	struct delayed_work	svc_resize_work;  /* resize svc_table */
+	atomic_t		svc_table_changes;/* ++ on new table */
 	/* Service counters */
 	atomic_t		num_services[IP_VS_AF_MAX];   /* Services */
 	atomic_t		fwm_services[IP_VS_AF_MAX];   /* Services */
@@ -1219,6 +1226,7 @@ struct netns_ipvs {
 	int			sysctl_est_nice;	/* kthread nice */
 	int			est_stopped;		/* stop tasks */
 #endif
+	int			sysctl_svc_lfactor;
 
 	/* ip_vs_lblc */
 	int			sysctl_lblc_expiration;
@@ -1228,6 +1236,7 @@ struct netns_ipvs {
 	int			sysctl_lblcr_expiration;
 	struct ctl_table_header	*lblcr_ctl_header;
 	struct ctl_table	*lblcr_ctl_table;
+	unsigned long		work_flags;	/* IP_VS_WORK_* flags */
 	/* ip_vs_est */
 	struct delayed_work	est_reload_work;/* Reload kthread tasks */
 	struct mutex		est_mutex;	/* protect kthread tasks */
@@ -1259,9 +1268,7 @@ struct netns_ipvs {
 	unsigned int		mixed_address_family_dests;
 	unsigned int		hooks_afmask;	/* &1=AF_INET, &2=AF_INET6 */
 
-	/* the service mutex that protect svc_table and svc_fwm_table */
-	struct mutex service_mutex;
-	struct hlist_head svc_table[IP_VS_SVC_TAB_SIZE];	/* Services */
+	struct ip_vs_rht __rcu	*svc_table;	/* Services */
 };
 
 #define DEFAULT_SYNC_THRESHOLD	3
@@ -1511,6 +1518,18 @@ static inline int sysctl_est_nice(struct netns_ipvs *ipvs)
 
 #endif
 
+/* Get load factor to map num_services/u_thresh to t->size
+ * Smaller value decreases u_thresh to reduce collisions but increases
+ * the table size
+ * Returns factor where:
+ * - <0: u_thresh = size >> -factor, eg. lfactor -2 = 25% load
+ * - >=0: u_thresh = size << factor, eg. lfactor 1 = 200% load
+ */
+static inline int sysctl_svc_lfactor(struct netns_ipvs *ipvs)
+{
+	return READ_ONCE(ipvs->sysctl_svc_lfactor);
+}
+
 /* IPVS core functions
  * (from ip_vs_core.c)
  */