net: ipv4: add support for ECMP hash policy choice

This patch adds support for ECMP hash policy choice via a new sysctl
called fib_multipath_hash_policy and also adds support for L4 hashes.
The current values for fib_multipath_hash_policy are:
 0 - layer 3 (default)
 1 - layer 4
If there's an skb hash already set and it matches the chosen policy then it
will be used instead of being calculated (currently only for L4).
In L3 mode we always calculate the hash due to the ICMP error special
case, the flow dissector's field consistentification should handle the
address order thus we can remove the address reversals.
If the skb is provided we always use it for the hash calculation,
otherwise we fallback to fl4, that is if skb is NULL fl4 has to be set.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Documentation/networking/ip-sysctl.txt

@@ -73,6 +73,14 @@ fib_multipath_use_neigh - BOOLEAN
 	0 - disabled
 	1 - enabled
 
+fib_multipath_hash_policy - INTEGER
+	Controls which hash policy to use for multipath routes. Only valid
+	for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
+	Default: 0 (Layer 3)
+	Possible values:
+	0 - Layer 3
+	1 - Layer 4
+
 route/max_size - INTEGER
 	Maximum number of routes allowed in the kernel.  Increase
 	this when using large numbers of interfaces and/or routes.

include/net/ip_fib.h

@@ -395,17 +395,13 @@ int fib_sync_down_dev(struct net_device *dev, unsigned long event, bool force);
 int fib_sync_down_addr(struct net_device *dev, __be32 local);
 int fib_sync_up(struct net_device *dev, unsigned int nh_flags);
 
-extern u32 fib_multipath_secret __read_mostly;
-
-static inline int fib_multipath_hash(__be32 saddr, __be32 daddr)
-{
-	return jhash_2words((__force u32)saddr, (__force u32)daddr,
-			    fib_multipath_secret) >> 1;
-}
-
+#ifdef CONFIG_IP_ROUTE_MULTIPATH
+int fib_multipath_hash(const struct fib_info *fi, const struct flowi4 *fl4,
+		       const struct sk_buff *skb);
+#endif
 void fib_select_multipath(struct fib_result *res, int hash);
 void fib_select_path(struct net *net, struct fib_result *res,
-		     struct flowi4 *fl4, int mp_hash);
+		     struct flowi4 *fl4, const struct sk_buff *skb);
 
 /* Exported by fib_trie.c */
 void fib_trie_init(void);

include/net/netns/ipv4.h

@@ -151,6 +151,7 @@ struct netns_ipv4 {
 #endif
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
 	int sysctl_fib_multipath_use_neigh;
+	int sysctl_fib_multipath_hash_policy;
 #endif
 
 	unsigned int	fib_seq;	/* protected by rtnl_mutex */

include/net/route.h

@@ -113,13 +113,13 @@ struct in_device;
 int ip_rt_init(void);
 void rt_cache_flush(struct net *net);
 void rt_flush_dev(struct net_device *dev);
-struct rtable *__ip_route_output_key_hash(struct net *, struct flowi4 *flp,
-					  int mp_hash);
+struct rtable *__ip_route_output_key_hash(struct net *net, struct flowi4 *flp,
+					  const struct sk_buff *skb);
 
 static inline struct rtable *__ip_route_output_key(struct net *net,
 						   struct flowi4 *flp)
 {
-	return __ip_route_output_key_hash(net, flp, -1);
+	return __ip_route_output_key_hash(net, flp, NULL);
 }
 
 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,

net/ipv4/fib_semantics.c

@@ -57,7 +57,6 @@ static unsigned int fib_info_cnt;
 static struct hlist_head fib_info_devhash[DEVINDEX_HASHSIZE];
 
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
-u32 fib_multipath_secret __read_mostly;
 
 #define for_nexthops(fi) {						\
 	int nhsel; const struct fib_nh *nh;				\
@@ -576,9 +575,6 @@ static void fib_rebalance(struct fib_info *fi)
 
 		atomic_set(&nexthop_nh->nh_upper_bound, upper_bound);
 	} endfor_nexthops(fi);
-
-	net_get_random_once(&fib_multipath_secret,
-			    sizeof(fib_multipath_secret));
 }
 
 static inline void fib_add_weight(struct fib_info *fi,
@@ -1641,7 +1637,7 @@ void fib_select_multipath(struct fib_result *res, int hash)
 #endif
 
 void fib_select_path(struct net *net, struct fib_result *res,
-		     struct flowi4 *fl4, int mp_hash)
+		     struct flowi4 *fl4, const struct sk_buff *skb)
 {
 	bool oif_check;
 
@@ -1650,10 +1646,9 @@ void fib_select_path(struct net *net, struct fib_result *res,
 
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
 	if (res->fi->fib_nhs > 1 && oif_check) {
-		if (mp_hash < 0)
-			mp_hash = get_hash_from_flowi4(fl4) >> 1;
+		int h = fib_multipath_hash(res->fi, fl4, skb);
 
-		fib_select_multipath(res, mp_hash);
+		fib_select_multipath(res, h);
 	}
 	else
 #endif

net/ipv4/icmp.c

@@ -464,22 +464,6 @@ out_bh_enable:
 	local_bh_enable();
 }
 
-#ifdef CONFIG_IP_ROUTE_MULTIPATH
-
-/* Source and destination is swapped. See ip_multipath_icmp_hash */
-static int icmp_multipath_hash_skb(const struct sk_buff *skb)
-{
-	const struct iphdr *iph = ip_hdr(skb);
-
-	return fib_multipath_hash(iph->daddr, iph->saddr);
-}
-
-#else
-
-#define icmp_multipath_hash_skb(skb) (-1)
-
-#endif
-
 static struct rtable *icmp_route_lookup(struct net *net,
 					struct flowi4 *fl4,
 					struct sk_buff *skb_in,
@@ -505,8 +489,7 @@ static struct rtable *icmp_route_lookup(struct net *net,
 	fl4->flowi4_oif = l3mdev_master_ifindex(skb_dst(skb_in)->dev);
 
 	security_skb_classify_flow(skb_in, flowi4_to_flowi(fl4));
-	rt = __ip_route_output_key_hash(net, fl4,
-					icmp_multipath_hash_skb(skb_in));
+	rt = __ip_route_output_key_hash(net, fl4, skb_in);
 	if (IS_ERR(rt))
 		return rt;
 

net/ipv4/route.c

@@ -1734,45 +1734,97 @@ out:
 }
 
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
-
 /* To make ICMP packets follow the right flow, the multipath hash is
- * calculated from the inner IP addresses in reverse order.
+ * calculated from the inner IP addresses.
  */
-static int ip_multipath_icmp_hash(struct sk_buff *skb)
+static void ip_multipath_l3_keys(const struct sk_buff *skb,
+				 struct flow_keys *hash_keys)
 {
 	const struct iphdr *outer_iph = ip_hdr(skb);
-	struct icmphdr _icmph;
+	const struct iphdr *inner_iph;
 	const struct icmphdr *icmph;
 	struct iphdr _inner_iph;
-	const struct iphdr *inner_iph;
+	struct icmphdr _icmph;
+
+	hash_keys->addrs.v4addrs.src = outer_iph->saddr;
+	hash_keys->addrs.v4addrs.dst = outer_iph->daddr;
+	if (likely(outer_iph->protocol != IPPROTO_ICMP))
+		return;
 
 	if (unlikely((outer_iph->frag_off & htons(IP_OFFSET)) != 0))
-		goto standard_hash;
+		return;
 
 	icmph = skb_header_pointer(skb, outer_iph->ihl * 4, sizeof(_icmph),
 				   &_icmph);
 	if (!icmph)
-		goto standard_hash;
+		return;
 
 	if (icmph->type != ICMP_DEST_UNREACH &&
 	    icmph->type != ICMP_REDIRECT &&
 	    icmph->type != ICMP_TIME_EXCEEDED &&
-	    icmph->type != ICMP_PARAMETERPROB) {
-		goto standard_hash;
-	}
+	    icmph->type != ICMP_PARAMETERPROB)
+		return;
 
 	inner_iph = skb_header_pointer(skb,
 				       outer_iph->ihl * 4 + sizeof(_icmph),
 				       sizeof(_inner_iph), &_inner_iph);
 	if (!inner_iph)
-		goto standard_hash;
-
-	return fib_multipath_hash(inner_iph->daddr, inner_iph->saddr);
-
-standard_hash:
-	return fib_multipath_hash(outer_iph->saddr, outer_iph->daddr);
+		return;
+	hash_keys->addrs.v4addrs.src = inner_iph->saddr;
+	hash_keys->addrs.v4addrs.dst = inner_iph->daddr;
 }
 
+/* if skb is set it will be used and fl4 can be NULL */
+int fib_multipath_hash(const struct fib_info *fi, const struct flowi4 *fl4,
+		       const struct sk_buff *skb)
+{
+	struct net *net = fi->fib_net;
+	struct flow_keys hash_keys;
+	u32 mhash;
+
+	switch (net->ipv4.sysctl_fib_multipath_hash_policy) {
+	case 0:
+		memset(&hash_keys, 0, sizeof(hash_keys));
+		hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+		if (skb) {
+			ip_multipath_l3_keys(skb, &hash_keys);
+		} else {
+			hash_keys.addrs.v4addrs.src = fl4->saddr;
+			hash_keys.addrs.v4addrs.dst = fl4->daddr;
+		}
+		break;
+	case 1:
+		/* skb is currently provided only when forwarding */
+		if (skb) {
+			unsigned int flag = FLOW_DISSECTOR_F_STOP_AT_ENCAP;
+			struct flow_keys keys;
+
+			/* short-circuit if we already have L4 hash present */
+			if (skb->l4_hash)
+				return skb_get_hash_raw(skb) >> 1;
+			memset(&hash_keys, 0, sizeof(hash_keys));
+			skb_flow_dissect_flow_keys(skb, &keys, flag);
+			hash_keys.addrs.v4addrs.src = keys.addrs.v4addrs.src;
+			hash_keys.addrs.v4addrs.dst = keys.addrs.v4addrs.dst;
+			hash_keys.ports.src = keys.ports.src;
+			hash_keys.ports.dst = keys.ports.dst;
+			hash_keys.basic.ip_proto = keys.basic.ip_proto;
+		} else {
+			memset(&hash_keys, 0, sizeof(hash_keys));
+			hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+			hash_keys.addrs.v4addrs.src = fl4->saddr;
+			hash_keys.addrs.v4addrs.dst = fl4->daddr;
+			hash_keys.ports.src = fl4->fl4_sport;
+			hash_keys.ports.dst = fl4->fl4_dport;
+			hash_keys.basic.ip_proto = fl4->flowi4_proto;
+		}
+		break;
+	}
+	mhash = flow_hash_from_keys(&hash_keys);
+
+	return mhash >> 1;
+}
+EXPORT_SYMBOL_GPL(fib_multipath_hash);
 #endif /* CONFIG_IP_ROUTE_MULTIPATH */
 
 static int ip_mkroute_input(struct sk_buff *skb,
@@ -1782,12 +1834,8 @@ static int ip_mkroute_input(struct sk_buff *skb,
 {
 #ifdef CONFIG_IP_ROUTE_MULTIPATH
 	if (res->fi && res->fi->fib_nhs > 1) {
-		int h;
+		int h = fib_multipath_hash(res->fi, NULL, skb);
 
-		if (unlikely(ip_hdr(skb)->protocol == IPPROTO_ICMP))
-			h = ip_multipath_icmp_hash(skb);
-		else
-			h = fib_multipath_hash(saddr, daddr);
 		fib_select_multipath(res, h);
 	}
 #endif
@@ -2203,7 +2251,7 @@ add:
  */
 
 struct rtable *__ip_route_output_key_hash(struct net *net, struct flowi4 *fl4,
-					  int mp_hash)
+					  const struct sk_buff *skb)
 {
 	struct net_device *dev_out = NULL;
 	__u8 tos = RT_FL_TOS(fl4);
@@ -2365,7 +2413,7 @@ struct rtable *__ip_route_output_key_hash(struct net *net, struct flowi4 *fl4,
 		goto make_route;
 	}
 
-	fib_select_path(net, &res, fl4, mp_hash);
+	fib_select_path(net, &res, fl4, skb);
 
 	dev_out = FIB_RES_DEV(res);
 	fl4->flowi4_oif = dev_out->ifindex;

net/ipv4/sysctl_net_ipv4.c

@@ -997,6 +997,15 @@ static struct ctl_table ipv4_net_table[] = {
 		.extra1		= &zero,
 		.extra2		= &one,
 	},
+	{
+		.procname	= "fib_multipath_hash_policy",
+		.data		= &init_net.ipv4.sysctl_fib_multipath_hash_policy,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &one,
+	},
 #endif
 	{
 		.procname	= "ip_unprivileged_port_start",