ipv6: lockless IPV6_FLOWINFO_SEND implementation
np->sndflow reads are racy.
Use one bit ftom atomic inet->inet_flags instead,
IPV6_FLOWINFO_SEND setsockopt() can be lockless.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
diff --git a/include/linux/ipv6.h b/include/linux/ipv6.h
index 10f521a..0925382 100644
--- a/include/linux/ipv6.h
+++ b/include/linux/ipv6.h
@@ -243,8 +243,7 @@ struct ipv6_pinfo {
} rxopt;
/* sockopt flags */
- __u8 sndflow:1,
- srcprefs:3; /* 001: prefer temporary address
+ __u8 srcprefs:3; /* 001: prefer temporary address
* 010: prefer public address
* 100: prefer care-of address
*/
diff --git a/include/net/inet_sock.h b/include/net/inet_sock.h
index befee0f..98e1195 100644
--- a/include/net/inet_sock.h
+++ b/include/net/inet_sock.h
@@ -277,6 +277,7 @@ enum {
INET_FLAGS_RECVERR6 = 26,
INET_FLAGS_REPFLOW = 27,
INET_FLAGS_RTALERT_ISOLATE = 28,
+ INET_FLAGS_SNDFLOW = 29,
};
/* cmsg flags for inet */
diff --git a/net/dccp/ipv6.c b/net/dccp/ipv6.c
index d7e63ee..4803f06 100644
--- a/net/dccp/ipv6.c
+++ b/net/dccp/ipv6.c
@@ -844,7 +844,7 @@ static int dccp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
memset(&fl6, 0, sizeof(fl6));
- if (np->sndflow) {
+ if (inet6_test_bit(SNDFLOW, sk)) {
fl6.flowlabel = usin->sin6_flowinfo & IPV6_FLOWINFO_MASK;
IP6_ECN_flow_init(fl6.flowlabel);
if (fl6.flowlabel & IPV6_FLOWLABEL_MASK) {
diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c
index bc01ad5..4dd809b 100644
--- a/net/ipv4/ping.c
+++ b/net/ipv4/ping.c
@@ -899,7 +899,6 @@ int ping_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int flags,
#if IS_ENABLED(CONFIG_IPV6)
} else if (family == AF_INET6) {
- struct ipv6_pinfo *np = inet6_sk(sk);
struct ipv6hdr *ip6 = ipv6_hdr(skb);
DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
@@ -908,7 +907,7 @@ int ping_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int flags,
sin6->sin6_port = 0;
sin6->sin6_addr = ip6->saddr;
sin6->sin6_flowinfo = 0;
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
sin6->sin6_flowinfo = ip6_flowinfo(ip6);
sin6->sin6_scope_id =
ipv6_iface_scope_id(&sin6->sin6_addr,
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 4873736..c6ad0d6 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -537,7 +537,7 @@ int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
}
sin->sin6_port = inet->inet_dport;
sin->sin6_addr = sk->sk_v6_daddr;
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
sin->sin6_flowinfo = np->flow_label;
BPF_CGROUP_RUN_SA_PROG(sk, (struct sockaddr *)sin,
CGROUP_INET6_GETPEERNAME);
diff --git a/net/ipv6/datagram.c b/net/ipv6/datagram.c
index 74673a5..cc6a502 100644
--- a/net/ipv6/datagram.c
+++ b/net/ipv6/datagram.c
@@ -80,7 +80,8 @@ int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr)
struct flowi6 fl6;
int err = 0;
- if (np->sndflow && (np->flow_label & IPV6_FLOWLABEL_MASK)) {
+ if (inet6_test_bit(SNDFLOW, sk) &&
+ (np->flow_label & IPV6_FLOWLABEL_MASK)) {
flowlabel = fl6_sock_lookup(sk, np->flow_label);
if (IS_ERR(flowlabel))
return -EINVAL;
@@ -163,7 +164,7 @@ int __ip6_datagram_connect(struct sock *sk, struct sockaddr *uaddr,
if (usin->sin6_family != AF_INET6)
return -EAFNOSUPPORT;
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
fl6_flowlabel = usin->sin6_flowinfo & IPV6_FLOWINFO_MASK;
if (ipv6_addr_any(&usin->sin6_addr)) {
@@ -491,7 +492,7 @@ int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len)
const struct ipv6hdr *ip6h = container_of((struct in6_addr *)(nh + serr->addr_offset),
struct ipv6hdr, daddr);
sin->sin6_addr = ip6h->daddr;
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
sin->sin6_flowinfo = ip6_flowinfo(ip6h);
sin->sin6_scope_id =
ipv6_iface_scope_id(&sin->sin6_addr,
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index 85ea426..e9dc6f8 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -500,6 +500,11 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
return -EINVAL;
WRITE_ONCE(np->pmtudisc, val);
return 0;
+ case IPV6_FLOWINFO_SEND:
+ if (optlen < sizeof(int))
+ return -EINVAL;
+ inet6_assign_bit(SNDFLOW, sk, valbool);
+ return 0;
}
if (needs_rtnl)
rtnl_lock();
@@ -948,12 +953,6 @@ int do_ipv6_setsockopt(struct sock *sk, int level, int optname,
goto e_inval;
retv = ip6_ra_control(sk, val);
break;
- case IPV6_FLOWINFO_SEND:
- if (optlen < sizeof(int))
- goto e_inval;
- np->sndflow = valbool;
- retv = 0;
- break;
case IPV6_FLOWLABEL_MGR:
retv = ipv6_flowlabel_opt(sk, optval, optlen);
break;
@@ -1381,7 +1380,7 @@ int do_ipv6_getsockopt(struct sock *sk, int level, int optname,
break;
case IPV6_FLOWINFO_SEND:
- val = np->sndflow;
+ val = inet6_test_bit(SNDFLOW, sk);
break;
case IPV6_FLOWLABEL_MGR:
diff --git a/net/ipv6/ping.c b/net/ipv6/ping.c
index 4444b61..e8fb0d2 100644
--- a/net/ipv6/ping.c
+++ b/net/ipv6/ping.c
@@ -89,7 +89,7 @@ static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
return -EAFNOSUPPORT;
}
daddr = &(u->sin6_addr);
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
fl6.flowlabel = u->sin6_flowinfo & IPV6_FLOWINFO_MASK;
if (__ipv6_addr_needs_scope_id(ipv6_addr_type(daddr)))
oif = u->sin6_scope_id;
diff --git a/net/ipv6/raw.c b/net/ipv6/raw.c
index 47372cc..a2aa54a 100644
--- a/net/ipv6/raw.c
+++ b/net/ipv6/raw.c
@@ -795,7 +795,7 @@ static int rawv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
return -EINVAL;
daddr = &sin6->sin6_addr;
- if (np->sndflow) {
+ if (inet6_test_bit(SNDFLOW, sk)) {
fl6.flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 201caf8..94afb8d 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -163,7 +163,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
memset(&fl6, 0, sizeof(fl6));
- if (np->sndflow) {
+ if (inet6_test_bit(SNDFLOW, sk)) {
fl6.flowlabel = usin->sin6_flowinfo&IPV6_FLOWINFO_MASK;
IP6_ECN_flow_init(fl6.flowlabel);
if (fl6.flowlabel&IPV6_FLOWLABEL_MASK) {
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index c17e19f..5e9312e 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -1429,7 +1429,7 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
fl6->fl6_dport = sin6->sin6_port;
daddr = &sin6->sin6_addr;
- if (np->sndflow) {
+ if (inet6_test_bit(SNDFLOW, sk)) {
fl6->flowlabel = sin6->sin6_flowinfo&IPV6_FLOWINFO_MASK;
if (fl6->flowlabel & IPV6_FLOWLABEL_MASK) {
flowlabel = fl6_sock_lookup(sk, fl6->flowlabel);
diff --git a/net/l2tp/l2tp_ip6.c b/net/l2tp/l2tp_ip6.c
index 40af243..44cfb72 100644
--- a/net/l2tp/l2tp_ip6.c
+++ b/net/l2tp/l2tp_ip6.c
@@ -431,7 +431,7 @@ static int l2tp_ip6_getname(struct socket *sock, struct sockaddr *uaddr,
return -ENOTCONN;
lsa->l2tp_conn_id = lsk->peer_conn_id;
lsa->l2tp_addr = sk->sk_v6_daddr;
- if (np->sndflow)
+ if (inet6_test_bit(SNDFLOW, sk))
lsa->l2tp_flowinfo = np->flow_label;
} else {
if (ipv6_addr_any(&sk->sk_v6_rcv_saddr))
@@ -529,7 +529,7 @@ static int l2tp_ip6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
return -EAFNOSUPPORT;
daddr = &lsa->l2tp_addr;
- if (np->sndflow) {
+ if (inet6_test_bit(SNDFLOW, sk)) {
fl6.flowlabel = lsa->l2tp_flowinfo & IPV6_FLOWINFO_MASK;
if (fl6.flowlabel & IPV6_FLOWLABEL_MASK) {
flowlabel = fl6_sock_lookup(sk, fl6.flowlabel);
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index 42b5b85..5c0ed59 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -296,7 +296,8 @@ static void sctp_v6_get_dst(struct sctp_transport *t, union sctp_addr *saddr,
if (t->flowlabel & SCTP_FLOWLABEL_SET_MASK)
fl6->flowlabel = htonl(t->flowlabel & SCTP_FLOWLABEL_VAL_MASK);
- if (np->sndflow && (fl6->flowlabel & IPV6_FLOWLABEL_MASK)) {
+ if (inet6_test_bit(SNDFLOW, sk) &&
+ (fl6->flowlabel & IPV6_FLOWLABEL_MASK)) {
struct ip6_flowlabel *flowlabel;
flowlabel = fl6_sock_lookup(sk, fl6->flowlabel);
