ipv6: lockless IPV6_ADDR_PREFERENCES implementation

We have data-races while reading np->srcprefs

Switch the field to a plain byte, add READ_ONCE()
and WRITE_ONCE() annotations where needed,
and IPV6_ADDR_PREFERENCES setsockopt() can now be lockless.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20230918142321.1794107-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Author: edumazet

include/linux/ipv6.h

@@ -243,7 +243,7 @@ struct ipv6_pinfo {
 	} rxopt;
 
 	/* sockopt flags */
-	__u8			srcprefs:3;	/* 001: prefer temporary address
+	__u8			srcprefs;	/* 001: prefer temporary address
 						 * 010: prefer public address
 						 * 100: prefer care-of address
 						 */

include/net/ip6_route.h

@@ -53,13 +53,12 @@ struct route_info {
  */
 static inline int rt6_srcprefs2flags(unsigned int srcprefs)
 {
-	/* No need to bitmask because srcprefs have only 3 bits. */
-	return srcprefs << 3;
+	return (srcprefs & IPV6_PREFER_SRC_MASK) << 3;
 }
 
 static inline unsigned int rt6_flags2srcprefs(int flags)
 {
-	return (flags >> 3) & 7;
+	return (flags >> 3) & IPV6_PREFER_SRC_MASK;
 }
 
 static inline bool rt6_need_strict(const struct in6_addr *daddr)

include/net/ipv6.h

@@ -1306,10 +1306,13 @@ static inline void ip6_sock_set_recverr(struct sock *sk)
 	inet6_set_bit(RECVERR6, sk);
 }
 
-static inline int __ip6_sock_set_addr_preferences(struct sock *sk, int val)
+#define IPV6_PREFER_SRC_MASK (IPV6_PREFER_SRC_TMP | IPV6_PREFER_SRC_PUBLIC | \
+			      IPV6_PREFER_SRC_COA)
+
+static inline int ip6_sock_set_addr_preferences(struct sock *sk, int val)
 {
+	unsigned int prefmask = ~IPV6_PREFER_SRC_MASK;
 	unsigned int pref = 0;
-	unsigned int prefmask = ~0;
 
 	/* check PUBLIC/TMP/PUBTMP_DEFAULT conflicts */
 	switch (val & (IPV6_PREFER_SRC_PUBLIC |
@@ -1359,20 +1362,11 @@ static inline int __ip6_sock_set_addr_preferences(struct sock *sk, int val)
 		return -EINVAL;
 	}
 
-	inet6_sk(sk)->srcprefs = (inet6_sk(sk)->srcprefs & prefmask) | pref;
+	WRITE_ONCE(inet6_sk(sk)->srcprefs,
+		   (READ_ONCE(inet6_sk(sk)->srcprefs) & prefmask) | pref);
 	return 0;
 }
 
-static inline int ip6_sock_set_addr_preferences(struct sock *sk, int val)
-{
-	int ret;
-
-	lock_sock(sk);
-	ret = __ip6_sock_set_addr_preferences(sk, val);
-	release_sock(sk);
-	return ret;
-}
-
 static inline void ip6_sock_set_recvpktinfo(struct sock *sk)
 {
 	lock_sock(sk);

net/ipv6/ip6_output.c

@@ -1113,7 +1113,7 @@ static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk,
 		rcu_read_lock();
 		from = rt ? rcu_dereference(rt->from) : NULL;
 		err = ip6_route_get_saddr(net, from, &fl6->daddr,
-					  sk ? inet6_sk(sk)->srcprefs : 0,
+					  sk ? READ_ONCE(inet6_sk(sk)->srcprefs) : 0,
 					  &fl6->saddr);
 		rcu_read_unlock();
 

net/ipv6/ipv6_sockglue.c

@@ -505,6 +505,10 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 			return -EINVAL;
 		inet6_assign_bit(SNDFLOW, sk, valbool);
 		return 0;
+	case IPV6_ADDR_PREFERENCES:
+		if (optlen < sizeof(int))
+			return -EINVAL;
+		return ip6_sock_set_addr_preferences(sk, val);
 	}
 	if (needs_rtnl)
 		rtnl_lock();
@@ -964,11 +968,6 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
 		retv = xfrm_user_policy(sk, optname, optval, optlen);
 		break;
 
-	case IPV6_ADDR_PREFERENCES:
-		if (optlen < sizeof(int))
-			goto e_inval;
-		retv = __ip6_sock_set_addr_preferences(sk, val);
-		break;
 	case IPV6_RECVFRAGSIZE:
 		np->rxopt.bits.recvfragsize = valbool;
 		retv = 0;
@@ -1415,23 +1414,25 @@ int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
 	}
 
 	case IPV6_ADDR_PREFERENCES:
+		{
+		u8 srcprefs = READ_ONCE(np->srcprefs);
 		val = 0;
 
-		if (np->srcprefs & IPV6_PREFER_SRC_TMP)
+		if (srcprefs & IPV6_PREFER_SRC_TMP)
 			val |= IPV6_PREFER_SRC_TMP;
-		else if (np->srcprefs & IPV6_PREFER_SRC_PUBLIC)
+		else if (srcprefs & IPV6_PREFER_SRC_PUBLIC)
 			val |= IPV6_PREFER_SRC_PUBLIC;
 		else {
 			/* XXX: should we return system default? */
 			val |= IPV6_PREFER_SRC_PUBTMP_DEFAULT;
 		}
 
-		if (np->srcprefs & IPV6_PREFER_SRC_COA)
+		if (srcprefs & IPV6_PREFER_SRC_COA)
 			val |= IPV6_PREFER_SRC_COA;
 		else
 			val |= IPV6_PREFER_SRC_HOME;
 		break;
-
+		}
 	case IPV6_MINHOPCOUNT:
 		val = READ_ONCE(np->min_hopcount);
 		break;

net/ipv6/route.c

@@ -2622,7 +2622,7 @@ static struct dst_entry *ip6_route_output_flags_noref(struct net *net,
 	if (!any_src)
 		flags |= RT6_LOOKUP_F_HAS_SADDR;
 	else if (sk)
-		flags |= rt6_srcprefs2flags(inet6_sk(sk)->srcprefs);
+		flags |= rt6_srcprefs2flags(READ_ONCE(inet6_sk(sk)->srcprefs));
 
 	return fib6_rule_lookup(net, fl6, NULL, flags, ip6_pol_route_output);
 }