Merge tag 'ipvs-for-v4.18' of http://git.kernel.org/pub/scm/linux/kernel/git/horms/ipvs-next

Simon Horman says:

====================
IPVS Updates for v4.18

please consider these IPVS enhancements for v4.18.

* Whitepace cleanup

* Add Maglev hashing algorithm as a IPVS scheduler

  Inju Song says "Implements the Google's Maglev hashing algorithm as a
  IPVS scheduler.  Basically it provides consistent hashing but offers some
  special features about disruption and load balancing.

  1) minimal disruption: when the set of destinations changes,
     a connection will likely be sent to the same destination
     as it was before.

  2) load balancing: each destination will receive an almost
     equal number of connections.

 Seel also: [3.4 Consistent Hasing] in
 https://www.usenix.org/system/files/conference/nsdi16/nsdi16-paper-eisenbud.pdf
 "

* Fix to correct implementation of Knuth's multiplicative hashing
  which is used in sh/dh/lblc/lblcr algorithms. Instead the
  implementation provided by the hash_32() macro is used.
====================

Signed-off-by: Pablo Neira Ayuso <[email protected]>

Author: ummakynes

include/net/ip_vs.h

@@ -668,6 +668,7 @@ struct ip_vs_dest {
 	volatile unsigned int	flags;		/* dest status flags */
 	atomic_t		conn_flags;	/* flags to copy to conn */
 	atomic_t		weight;		/* server weight */
+	atomic_t		last_weight;	/* server latest weight */
 
 	refcount_t		refcnt;		/* reference counter */
 	struct ip_vs_stats      stats;          /* statistics */

net/netfilter/ipvs/Kconfig

@@ -225,6 +225,25 @@ config	IP_VS_SH
 	  If you want to compile it in kernel, say Y. To compile it as a
 	  module, choose M here. If unsure, say N.
 
+config	IP_VS_MH
+	tristate "maglev hashing scheduling"
+	---help---
+	  The maglev consistent hashing scheduling algorithm provides the
+	  Google's Maglev hashing algorithm as a IPVS scheduler. It assigns
+	  network connections to the servers through looking up a statically
+	  assigned special hash table called the lookup table. Maglev hashing
+	  is to assign a preference list of all the lookup table positions
+	  to each destination.
+
+	  Through this operation, The maglev hashing gives an almost equal
+	  share of the lookup table to each of the destinations and provides
+	  minimal disruption by using the lookup table. When the set of
+	  destinations changes, a connection will likely be sent to the same
+	  destination as it was before.
+
+	  If you want to compile it in kernel, say Y. To compile it as a
+	  module, choose M here. If unsure, say N.
+
 config	IP_VS_SED
 	tristate "shortest expected delay scheduling"
 	---help---
@@ -266,6 +285,24 @@ config IP_VS_SH_TAB_BITS
 	  needs to be large enough to effectively fit all the destinations
 	  multiplied by their respective weights.
 
+comment 'IPVS MH scheduler'
+
+config IP_VS_MH_TAB_INDEX
+	int "IPVS maglev hashing table index of size (the prime numbers)"
+	range 8 17
+	default 12
+	---help---
+	  The maglev hashing scheduler maps source IPs to destinations
+	  stored in a hash table. This table is assigned by a preference
+	  list of the positions to each destination until all slots in
+	  the table are filled. The index determines the prime for size of
+	  the table as�251, 509, 1021, 2039, 4093, 8191, 16381, 32749,
+	  65521 or 131071.�When using weights to allow destinations to
+	  receive more connections,�the table is assigned an amount
+	  proportional to the weights specified.�The table needs to be large
+	  enough to effectively fit all the destinations multiplied by their
+	  respective weights.
+
 comment 'IPVS application helper'
 
 config	IP_VS_FTP

net/netfilter/ipvs/Makefile

@@ -33,6 +33,7 @@ obj-$(CONFIG_IP_VS_LBLC) += ip_vs_lblc.o
 obj-$(CONFIG_IP_VS_LBLCR) += ip_vs_lblcr.o
 obj-$(CONFIG_IP_VS_DH) += ip_vs_dh.o
 obj-$(CONFIG_IP_VS_SH) += ip_vs_sh.o
+obj-$(CONFIG_IP_VS_MH) += ip_vs_mh.o
 obj-$(CONFIG_IP_VS_SED) += ip_vs_sed.o
 obj-$(CONFIG_IP_VS_NQ) += ip_vs_nq.o
 

net/netfilter/ipvs/ip_vs_ctl.c

@@ -821,6 +821,10 @@ __ip_vs_update_dest(struct ip_vs_service *svc, struct ip_vs_dest *dest,
 	if (add && udest->af != svc->af)
 		ipvs->mixed_address_family_dests++;
 
+	/* keep the last_weight with latest non-0 weight */
+	if (add || udest->weight != 0)
+		atomic_set(&dest->last_weight, udest->weight);
+
 	/* set the weight and the flags */
 	atomic_set(&dest->weight, udest->weight);
 	conn_flags = udest->conn_flags & IP_VS_CONN_F_DEST_MASK;

net/netfilter/ipvs/ip_vs_dh.c

@@ -43,6 +43,7 @@
 #include <linux/module.h>
 #include <linux/kernel.h>
 #include <linux/skbuff.h>
+#include <linux/hash.h>
 
 #include <net/ip_vs.h>
 
@@ -81,7 +82,7 @@ static inline unsigned int ip_vs_dh_hashkey(int af, const union nf_inet_addr *ad
 		addr_fold = addr->ip6[0]^addr->ip6[1]^
 			    addr->ip6[2]^addr->ip6[3];
 #endif
-	return (ntohl(addr_fold)*2654435761UL) & IP_VS_DH_TAB_MASK;
+	return hash_32(ntohl(addr_fold), IP_VS_DH_TAB_BITS);
 }
 
 

net/netfilter/ipvs/ip_vs_lblc.c

@@ -48,6 +48,7 @@
 #include <linux/kernel.h>
 #include <linux/skbuff.h>
 #include <linux/jiffies.h>
+#include <linux/hash.h>
 
 /* for sysctl */
 #include <linux/fs.h>
@@ -160,7 +161,7 @@ ip_vs_lblc_hashkey(int af, const union nf_inet_addr *addr)
 		addr_fold = addr->ip6[0]^addr->ip6[1]^
 			    addr->ip6[2]^addr->ip6[3];
 #endif
-	return (ntohl(addr_fold)*2654435761UL) & IP_VS_LBLC_TAB_MASK;
+	return hash_32(ntohl(addr_fold), IP_VS_LBLC_TAB_BITS);
 }
 
 

net/netfilter/ipvs/ip_vs_lblcr.c

@@ -47,6 +47,7 @@
 #include <linux/jiffies.h>
 #include <linux/list.h>
 #include <linux/slab.h>
+#include <linux/hash.h>
 
 /* for sysctl */
 #include <linux/fs.h>
@@ -323,7 +324,7 @@ ip_vs_lblcr_hashkey(int af, const union nf_inet_addr *addr)
 		addr_fold = addr->ip6[0]^addr->ip6[1]^
 			    addr->ip6[2]^addr->ip6[3];
 #endif
-	return (ntohl(addr_fold)*2654435761UL) & IP_VS_LBLCR_TAB_MASK;
+	return hash_32(ntohl(addr_fold), IP_VS_LBLCR_TAB_BITS);
 }