inet: add RCU protection to inet->opt

We lack proper synchronization to manipulate inet->opt ip_options

Problem is ip_make_skb() calls ip_setup_cork() and
ip_setup_cork() possibly makes a copy of ipc->opt (struct ip_options),
without any protection against another thread manipulating inet->opt.

Another thread can change inet->opt pointer and free old one under us.

Use RCU to protect inet->opt (changed to inet->inet_opt).

Instead of handling atomic refcounts, just copy ip_options when
necessary, to avoid cache line dirtying.

We cant insert an rcu_head in struct ip_options since its included in
skb->cb[], so this patch is large because I had to introduce a new
ip_options_rcu structure.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

11 files changed, 205 insertions, 149 deletions

diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 0413af3..963a621 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -153,7 +153,7 @@ void inet_sock_destruct(struct sock *sk)
 	WARN_ON(sk->sk_wmem_queued);
 	WARN_ON(sk->sk_forward_alloc);
 
-	kfree(inet->opt);
+	kfree(rcu_dereference_protected(inet->inet_opt, 1));
 	dst_release(rcu_dereference_check(sk->sk_dst_cache, 1));
 	sk_refcnt_debug_dec(sk);
 }
@@ -1106,9 +1106,12 @@ static int inet_sk_reselect_saddr(struct sock *sk)
 	struct flowi4 fl4;
 	struct rtable *rt;
 	__be32 new_saddr;
+	struct ip_options_rcu *inet_opt;
 
-	if (inet->opt && inet->opt->srr)
-		daddr = inet->opt->faddr;
+	inet_opt = rcu_dereference_protected(inet->inet_opt,
+					     sock_owned_by_user(sk));
+	if (inet_opt && inet_opt->opt.srr)
+		daddr = inet_opt->opt.faddr;
 
 	/* Query new route. */
 	rt = ip_route_connect(&fl4, daddr, 0, RT_CONN_FLAGS(sk),
@@ -1148,6 +1151,7 @@ int inet_sk_rebuild_header(struct sock *sk)
 	struct inet_sock *inet = inet_sk(sk);
 	struct rtable *rt = (struct rtable *)__sk_dst_check(sk, 0);
 	__be32 daddr;
+	struct ip_options_rcu *inet_opt;
 	int err;
 
 	/* Route is OK, nothing to do. */
@@ -1155,9 +1159,12 @@ int inet_sk_rebuild_header(struct sock *sk)
 		return 0;
 
 	/* Reroute. */
+	rcu_read_lock();
+	inet_opt = rcu_dereference(inet->inet_opt);
 	daddr = inet->inet_daddr;
-	if (inet->opt && inet->opt->srr)
-		daddr = inet->opt->faddr;
+	if (inet_opt && inet_opt->opt.srr)
+		daddr = inet_opt->opt.faddr;
+	rcu_read_unlock();
 	rt = ip_route_output_ports(sock_net(sk), sk, daddr, inet->inet_saddr,
 				   inet->inet_dport, inet->inet_sport,
 				   sk->sk_protocol, RT_CONN_FLAGS(sk),
diff --git a/net/ipv4/cipso_ipv4.c b/net/ipv4/cipso_ipv4.c
index a0af7ea..2b3c23c 100644
--- a/net/ipv4/cipso_ipv4.c
+++ b/net/ipv4/cipso_ipv4.c
@@ -1857,6 +1857,11 @@ static int cipso_v4_genopt(unsigned char *buf, u32 buf_len,
 	return CIPSO_V4_HDR_LEN + ret_val;
 }
 
+static void opt_kfree_rcu(struct rcu_head *head)
+{
+	kfree(container_of(head, struct ip_options_rcu, rcu));
+}
+
 /**
  * cipso_v4_sock_setattr - Add a CIPSO option to a socket
  * @sk: the socket
@@ -1879,7 +1884,7 @@ int cipso_v4_sock_setattr(struct sock *sk,
 	unsigned char *buf = NULL;
 	u32 buf_len;
 	u32 opt_len;
-	struct ip_options *opt = NULL;
+	struct ip_options_rcu *old, *opt = NULL;
 	struct inet_sock *sk_inet;
 	struct inet_connection_sock *sk_conn;
 
@@ -1915,22 +1920,25 @@ int cipso_v4_sock_setattr(struct sock *sk,
 		ret_val = -ENOMEM;
 		goto socket_setattr_failure;
 	}
-	memcpy(opt->__data, buf, buf_len);
-	opt->optlen = opt_len;
-	opt->cipso = sizeof(struct iphdr);
+	memcpy(opt->opt.__data, buf, buf_len);
+	opt->opt.optlen = opt_len;
+	opt->opt.cipso = sizeof(struct iphdr);
 	kfree(buf);
 	buf = NULL;
 
 	sk_inet = inet_sk(sk);
+
+	old = rcu_dereference_protected(sk_inet->inet_opt, sock_owned_by_user(sk));
 	if (sk_inet->is_icsk) {
 		sk_conn = inet_csk(sk);
-		if (sk_inet->opt)
-			sk_conn->icsk_ext_hdr_len -= sk_inet->opt->optlen;
-		sk_conn->icsk_ext_hdr_len += opt->optlen;
+		if (old)
+			sk_conn->icsk_ext_hdr_len -= old->opt.optlen;
+		sk_conn->icsk_ext_hdr_len += opt->opt.optlen;
 		sk_conn->icsk_sync_mss(sk, sk_conn->icsk_pmtu_cookie);
 	}
-	opt = xchg(&sk_inet->opt, opt);
-	kfree(opt);
+	rcu_assign_pointer(sk_inet->inet_opt, opt);
+	if (old)
+		call_rcu(&old->rcu, opt_kfree_rcu);
 
 	return 0;
 
@@ -1960,7 +1968,7 @@ int cipso_v4_req_setattr(struct request_sock *req,
 	unsigned char *buf = NULL;
 	u32 buf_len;
 	u32 opt_len;
-	struct ip_options *opt = NULL;
+	struct ip_options_rcu *opt = NULL;
 	struct inet_request_sock *req_inet;
 
 	/* We allocate the maximum CIPSO option size here so we are probably
@@ -1988,15 +1996,16 @@ int cipso_v4_req_setattr(struct request_sock *req,
 		ret_val = -ENOMEM;
 		goto req_setattr_failure;
 	}
-	memcpy(opt->__data, buf, buf_len);
-	opt->optlen = opt_len;
-	opt->cipso = sizeof(struct iphdr);
+	memcpy(opt->opt.__data, buf, buf_len);
+	opt->opt.optlen = opt_len;
+	opt->opt.cipso = sizeof(struct iphdr);
 	kfree(buf);
 	buf = NULL;
 
 	req_inet = inet_rsk(req);
 	opt = xchg(&req_inet->opt, opt);
-	kfree(opt);
+	if (opt)
+		call_rcu(&opt->rcu, opt_kfree_rcu);
 
 	return 0;
 
@@ -2016,34 +2025,34 @@ req_setattr_failure:
  * values on failure.
  *
  */
-static int cipso_v4_delopt(struct ip_options **opt_ptr)
+static int cipso_v4_delopt(struct ip_options_rcu **opt_ptr)
 {
 	int hdr_delta = 0;
-	struct ip_options *opt = *opt_ptr;
+	struct ip_options_rcu *opt = *opt_ptr;
 
-	if (opt->srr || opt->rr || opt->ts || opt->router_alert) {
+	if (opt->opt.srr || opt->opt.rr || opt->opt.ts || opt->opt.router_alert) {
 		u8 cipso_len;
 		u8 cipso_off;
 		unsigned char *cipso_ptr;
 		int iter;
 		int optlen_new;
 
-		cipso_off = opt->cipso - sizeof(struct iphdr);
-		cipso_ptr = &opt->__data[cipso_off];
+		cipso_off = opt->opt.cipso - sizeof(struct iphdr);
+		cipso_ptr = &opt->opt.__data[cipso_off];
 		cipso_len = cipso_ptr[1];
 
-		if (opt->srr > opt->cipso)
-			opt->srr -= cipso_len;
-		if (opt->rr > opt->cipso)
-			opt->rr -= cipso_len;
-		if (opt->ts > opt->cipso)
-			opt->ts -= cipso_len;
-		if (opt->router_alert > opt->cipso)
-			opt->router_alert -= cipso_len;
-		opt->cipso = 0;
+		if (opt->opt.srr > opt->opt.cipso)
+			opt->opt.srr -= cipso_len;
+		if (opt->opt.rr > opt->opt.cipso)
+			opt->opt.rr -= cipso_len;
+		if (opt->opt.ts > opt->opt.cipso)
+			opt->opt.ts -= cipso_len;
+		if (opt->opt.router_alert > opt->opt.cipso)
+			opt->opt.router_alert -= cipso_len;
+		opt->opt.cipso = 0;
 
 		memmove(cipso_ptr, cipso_ptr + cipso_len,
-			opt->optlen - cipso_off - cipso_len);
+			opt->opt.optlen - cipso_off - cipso_len);
 
 		/* determining the new total option length is tricky because of
 		 * the padding necessary, the only thing i can think to do at
@@ -2052,21 +2061,21 @@ static int cipso_v4_delopt(struct ip_options **opt_ptr)
 		 * from there we can determine the new total option length */
 		iter = 0;
 		optlen_new = 0;
-		while (iter < opt->optlen)
-			if (opt->__data[iter] != IPOPT_NOP) {
-				iter += opt->__data[iter + 1];
+		while (iter < opt->opt.optlen)
+			if (opt->opt.__data[iter] != IPOPT_NOP) {
+				iter += opt->opt.__data[iter + 1];
 				optlen_new = iter;
 			} else
 				iter++;
-		hdr_delta = opt->optlen;
-		opt->optlen = (optlen_new + 3) & ~3;
-		hdr_delta -= opt->optlen;
+		hdr_delta = opt->opt.optlen;
+		opt->opt.optlen = (optlen_new + 3) & ~3;
+		hdr_delta -= opt->opt.optlen;
 	} else {
 		/* only the cipso option was present on the socket so we can
 		 * remove the entire option struct */
 		*opt_ptr = NULL;
-		hdr_delta = opt->optlen;
-		kfree(opt);
+		hdr_delta = opt->opt.optlen;
+		call_rcu(&opt->rcu, opt_kfree_rcu);
 	}
 
 	return hdr_delta;
@@ -2083,15 +2092,15 @@ static int cipso_v4_delopt(struct ip_options **opt_ptr)
 void cipso_v4_sock_delattr(struct sock *sk)
 {
 	int hdr_delta;
-	struct ip_options *opt;
+	struct ip_options_rcu *opt;
 	struct inet_sock *sk_inet;
 
 	sk_inet = inet_sk(sk);
-	opt = sk_inet->opt;
-	if (opt == NULL || opt->cipso == 0)
+	opt = rcu_dereference_protected(sk_inet->inet_opt, 1);
+	if (opt == NULL || opt->opt.cipso == 0)
 		return;
 
-	hdr_delta = cipso_v4_delopt(&sk_inet->opt);
+	hdr_delta = cipso_v4_delopt(&sk_inet->inet_opt);
 	if (sk_inet->is_icsk && hdr_delta > 0) {
 		struct inet_connection_sock *sk_conn = inet_csk(sk);
 		sk_conn->icsk_ext_hdr_len -= hdr_delta;
@@ -2109,12 +2118,12 @@ void cipso_v4_sock_delattr(struct sock *sk)
  */
 void cipso_v4_req_delattr(struct request_sock *req)
 {
-	struct ip_options *opt;
+	struct ip_options_rcu *opt;
 	struct inet_request_sock *req_inet;
 
 	req_inet = inet_rsk(req);
 	opt = req_inet->opt;
-	if (opt == NULL || opt->cipso == 0)
+	if (opt == NULL || opt->opt.cipso == 0)
 		return;
 
 	cipso_v4_delopt(&req_inet->opt);
@@ -2184,14 +2193,18 @@ getattr_return:
  */
 int cipso_v4_sock_getattr(struct sock *sk, struct netlbl_lsm_secattr *secattr)
 {
-	struct ip_options *opt;
+	struct ip_options_rcu *opt;
+	int res = -ENOMSG;
 
-	opt = inet_sk(sk)->opt;
-	if (opt == NULL || opt->cipso == 0)
-		return -ENOMSG;
-
-	return cipso_v4_getattr(opt->__data + opt->cipso - sizeof(struct iphdr),
-				secattr);
+	rcu_read_lock();
+	opt = rcu_dereference(inet_sk(sk)->inet_opt);
+	if (opt && opt->opt.cipso)
+		res = cipso_v4_getattr(opt->opt.__data +
+						opt->opt.cipso -
+						sizeof(struct iphdr),
+				       secattr);
+	rcu_read_unlock();
+	return res;
 }
 
 /**
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index 74e35e5..cfeca3c 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -108,8 +108,7 @@ struct icmp_bxm {
 		__be32	       times[3];
 	} data;
 	int head_len;
-	struct ip_options replyopts;
-	unsigned char  optbuf[40];
+	struct ip_options_data replyopts;
 };
 
 /* An array of errno for error messages from dest unreach. */
@@ -333,7 +332,7 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
 	struct inet_sock *inet;
 	__be32 daddr;
 
-	if (ip_options_echo(&icmp_param->replyopts, skb))
+	if (ip_options_echo(&icmp_param->replyopts.opt.opt, skb))
 		return;
 
 	sk = icmp_xmit_lock(net);
@@ -347,10 +346,10 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
 	daddr = ipc.addr = rt->rt_src;
 	ipc.opt = NULL;
 	ipc.tx_flags = 0;
-	if (icmp_param->replyopts.optlen) {
-		ipc.opt = &icmp_param->replyopts;
-		if (ipc.opt->srr)
-			daddr = icmp_param->replyopts.faddr;
+	if (icmp_param->replyopts.opt.opt.optlen) {
+		ipc.opt = &icmp_param->replyopts.opt;
+		if (ipc.opt->opt.srr)
+			daddr = icmp_param->replyopts.opt.opt.faddr;
 	}
 	{
 		struct flowi4 fl4 = {
@@ -379,8 +378,8 @@ static struct rtable *icmp_route_lookup(struct net *net, struct sk_buff *skb_in,
 					struct icmp_bxm *param)
 {
 	struct flowi4 fl4 = {
-		.daddr = (param->replyopts.srr ?
-			  param->replyopts.faddr : iph->saddr),
+		.daddr = (param->replyopts.opt.opt.srr ?
+			  param->replyopts.opt.opt.faddr : iph->saddr),
 		.saddr = saddr,
 		.flowi4_tos = RT_TOS(tos),
 		.flowi4_proto = IPPROTO_ICMP,
@@ -581,7 +580,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
 					   IPTOS_PREC_INTERNETCONTROL) :
 					  iph->tos;
 
-	if (ip_options_echo(&icmp_param.replyopts, skb_in))
+	if (ip_options_echo(&icmp_param.replyopts.opt.opt, skb_in))
 		goto out_unlock;
 
 
@@ -597,7 +596,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
 	icmp_param.offset = skb_network_offset(skb_in);
 	inet_sk(sk)->tos = tos;
 	ipc.addr = iph->saddr;
-	ipc.opt = &icmp_param.replyopts;
+	ipc.opt = &icmp_param.replyopts.opt;
 	ipc.tx_flags = 0;
 
 	rt = icmp_route_lookup(net, skb_in, iph, saddr, tos,
@@ -613,7 +612,7 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, __be32 info)
 	room = dst_mtu(&rt->dst);
 	if (room > 576)
 		room = 576;
-	room -= sizeof(struct iphdr) + icmp_param.replyopts.optlen;
+	room -= sizeof(struct iphdr) + icmp_param.replyopts.opt.opt.optlen;
 	room -= sizeof(struct icmphdr);
 
 	icmp_param.data_len = skb_in->len - icmp_param.offset;
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 8514db5..3282cb2 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -354,20 +354,20 @@ struct dst_entry *inet_csk_route_req(struct sock *sk,
 {
 	struct rtable *rt;
 	const struct inet_request_sock *ireq = inet_rsk(req);
-	struct ip_options *opt = inet_rsk(req)->opt;
+	struct ip_options_rcu *opt = inet_rsk(req)->opt;
 	struct net *net = sock_net(sk);
 	struct flowi4 fl4;
 
 	flowi4_init_output(&fl4, sk->sk_bound_dev_if, sk->sk_mark,
 			   RT_CONN_FLAGS(sk), RT_SCOPE_UNIVERSE,
 			   sk->sk_protocol, inet_sk_flowi_flags(sk),
-			   (opt && opt->srr) ? opt->faddr : ireq->rmt_addr,
+			   (opt && opt->opt.srr) ? opt->opt.faddr : ireq->rmt_addr,
 			   ireq->loc_addr, ireq->rmt_port, inet_sk(sk)->inet_sport);
 	security_req_classify_flow(req, flowi4_to_flowi(&fl4));
 	rt = ip_route_output_flow(net, &fl4, sk);
 	if (IS_ERR(rt))
 		goto no_route;
-	if (opt && opt->is_strictroute && rt->rt_dst != rt->rt_gateway)
+	if (opt && opt->opt.is_strictroute && rt->rt_dst != rt->rt_gateway)
 		goto route_err;
 	return &rt->dst;
 
diff --git a/net/ipv4/ip_options.c b/net/ipv4/ip_options.c
index 2391b24..01fc409 100644
--- a/net/ipv4/ip_options.c
+++ b/net/ipv4/ip_options.c
@@ -36,7 +36,7 @@
  * saddr is address of outgoing interface.
  */
 
-void ip_options_build(struct sk_buff * skb, struct ip_options * opt,
+void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
 			    __be32 daddr, struct rtable *rt, int is_frag)
 {
 	unsigned char *iph = skb_network_header(skb);
@@ -83,9 +83,9 @@ void ip_options_build(struct sk_buff * skb, struct ip_options * opt,
  * NOTE: dopt cannot point to skb.
  */
 
-int ip_options_echo(struct ip_options * dopt, struct sk_buff * skb)
+int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb)
 {
-	struct ip_options *sopt;
+	const struct ip_options *sopt;
 	unsigned char *sptr, *dptr;
 	int soffset, doffset;
 	int	optlen;
@@ -95,10 +95,8 @@ int ip_options_echo(struct ip_options * dopt, struct sk_buff * skb)
 
 	sopt = &(IPCB(skb)->opt);
 
-	if (sopt->optlen == 0) {
-		dopt->optlen = 0;
+	if (sopt->optlen == 0)
 		return 0;
-	}
 
 	sptr = skb_network_header(skb);
 	dptr = dopt->__data;
@@ -157,7 +155,7 @@ int ip_options_echo(struct ip_options * dopt, struct sk_buff * skb)
 		dopt->optlen += optlen;
 	}
 	if (sopt->srr) {
-		unsigned char * start = sptr+sopt->srr;
+		unsigned char *start = sptr+sopt->srr;
 		__be32 faddr;
 
 		optlen  = start[1];
@@ -499,19 +497,19 @@ void ip_options_undo(struct ip_options * opt)
 	}
 }
 
-static struct ip_options *ip_options_get_alloc(const int optlen)
+static struct ip_options_rcu *ip_options_get_alloc(const int optlen)
 {
-	return kzalloc(sizeof(struct ip_options) + ((optlen + 3) & ~3),
+	return kzalloc(sizeof(struct ip_options_rcu) + ((optlen + 3) & ~3),
 		       GFP_KERNEL);
 }
 
-static int ip_options_get_finish(struct net *net, struct ip_options **optp,
-				 struct ip_options *opt, int optlen)
+static int ip_options_get_finish(struct net *net, struct ip_options_rcu **optp,
+				 struct ip_options_rcu *opt, int optlen)
 {
 	while (optlen & 3)
-		opt->__data[optlen++] = IPOPT_END;
-	opt->optlen = optlen;
-	if (optlen && ip_options_compile(net, opt, NULL)) {
+		opt->opt.__data[optlen++] = IPOPT_END;
+	opt->opt.optlen = optlen;
+	if (optlen && ip_options_compile(net, &opt->opt, NULL)) {
 		kfree(opt);
 		return -EINVAL;
 	}
@@ -520,29 +518,29 @@ static int ip_options_get_finish(struct net *net, struct ip_options **optp,
 	return 0;
 }
 
-int ip_options_get_from_user(struct net *net, struct ip_options **optp,
+int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
 			     unsigned char __user *data, int optlen)
 {
-	struct ip_options *opt = ip_options_get_alloc(optlen);
+	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);
 
 	if (!opt)
 		return -ENOMEM;
-	if (optlen && copy_from_user(opt->__data, data, optlen)) {
+	if (optlen && copy_from_user(opt->opt.__data, data, optlen)) {
 		kfree(opt);
 		return -EFAULT;
 	}
 	return ip_options_get_finish(net, optp, opt, optlen);
 }
 
-int ip_options_get(struct net *net, struct ip_options **optp,
+int ip_options_get(struct net *net, struct ip_options_rcu **optp,
 		   unsigned char *data, int optlen)
 {
-	struct ip_options *opt = ip_options_get_alloc(optlen);
+	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);
 
 	if (!opt)
 		return -ENOMEM;
 	if (optlen)
-		memcpy(opt->__data, data, optlen);
+		memcpy(opt->opt.__data, data, optlen);
 	return ip_options_get_finish(net, optp, opt, optlen);
 }
 
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
index bdad3d6..362e66f 100644
--- a/net/ipv4/ip_output.c
+++ b/net/ipv4/ip_output.c
@@ -140,14 +140,14 @@ static inline int ip_select_ttl(struct inet_sock *inet, struct dst_entry *dst)
  *
  */
 int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
-			  __be32 saddr, __be32 daddr, struct ip_options *opt)
+			  __be32 saddr, __be32 daddr, struct ip_options_rcu *opt)
 {
 	struct inet_sock *inet = inet_sk(sk);
 	struct rtable *rt = skb_rtable(skb);
 	struct iphdr *iph;
 
 	/* Build the IP header. */
-	skb_push(skb, sizeof(struct iphdr) + (opt ? opt->optlen : 0));
+	skb_push(skb, sizeof(struct iphdr) + (opt ? opt->opt.optlen : 0));
 	skb_reset_network_header(skb);
 	iph = ip_hdr(skb);
 	iph->version  = 4;
@@ -163,9 +163,9 @@ int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
 	iph->protocol = sk->sk_protocol;
 	ip_select_ident(iph, &rt->dst, sk);
 
-	if (opt && opt->optlen) {
-		iph->ihl += opt->optlen>>2;
-		ip_options_build(skb, opt, daddr, rt, 0);
+	if (opt && opt->opt.optlen) {
+		iph->ihl += opt->opt.optlen>>2;
+		ip_options_build(skb, &opt->opt, daddr, rt, 0);
 	}
 
 	skb->priority = sk->sk_priority;
@@ -316,7 +316,7 @@ int ip_queue_xmit(struct sk_buff *skb)
 {
 	struct sock *sk = skb->sk;
 	struct inet_sock *inet = inet_sk(sk);
-	struct ip_options *opt = inet->opt;
+	struct ip_options_rcu *inet_opt;
 	struct rtable *rt;
 	struct iphdr *iph;
 	int res;
@@ -325,6 +325,7 @@ int ip_queue_xmit(struct sk_buff *skb)
 	 * f.e. by something like SCTP.
 	 */
 	rcu_read_lock();
+	inet_opt = rcu_dereference(inet->inet_opt);
 	rt = skb_rtable(skb);
 	if (rt != NULL)
 		goto packet_routed;
@@ -336,8 +337,8 @@ int ip_queue_xmit(struct sk_buff *skb)
 
 		/* Use correct destination address if we have options. */
 		daddr = inet->inet_daddr;
-		if(opt && opt->srr)
-			daddr = opt->faddr;
+		if (inet_opt && inet_opt->opt.srr)
+			daddr = inet_opt->opt.faddr;
 
 		/* If this fails, retransmit mechanism of transport layer will
 		 * keep trying until route appears or the connection times
@@ -357,11 +358,11 @@ int ip_queue_xmit(struct sk_buff *skb)
 	skb_dst_set_noref(skb, &rt->dst);
 
 packet_routed:
-	if (opt && opt->is_strictroute && rt->rt_dst != rt->rt_gateway)
+	if (inet_opt && inet_opt->opt.is_strictroute && rt->rt_dst != rt->rt_gateway)
 		goto no_route;
 
 	/* OK, we know where to send it, allocate and build IP header. */
-	skb_push(skb, sizeof(struct iphdr) + (opt ? opt->optlen : 0));
+	skb_push(skb, sizeof(struct iphdr) + (inet_opt ? inet_opt->opt.optlen : 0));
 	skb_reset_network_header(skb);
 	iph = ip_hdr(skb);
 	*((__be16 *)iph) = htons((4 << 12) | (5 << 8) | (inet->tos & 0xff));
@@ -375,9 +376,9 @@ packet_routed:
 	iph->daddr    = rt->rt_dst;
 	/* Transport layer set skb->h.foo itself. */
 
-	if (opt && opt->optlen) {
-		iph->ihl += opt->optlen >> 2;
-		ip_options_build(skb, opt, inet->inet_daddr, rt, 0);
+	if (inet_opt && inet_opt->opt.optlen) {
+		iph->ihl += inet_opt->opt.optlen >> 2;
+		ip_options_build(skb, &inet_opt->opt, inet->inet_daddr, rt, 0);
 	}
 
 	ip_select_ident_more(iph, &rt->dst, sk,
@@ -1033,7 +1034,7 @@ static int ip_setup_cork(struct sock *sk, struct inet_cork *cork,
 			 struct ipcm_cookie *ipc, struct rtable **rtp)
 {
 	struct inet_sock *inet = inet_sk(sk);
-	struct ip_options *opt;
+	struct ip_options_rcu *opt;
 	struct rtable *rt;
 
 	/*
@@ -1047,7 +1048,7 @@ static int ip_setup_cork(struct sock *sk, struct inet_cork *cork,
 			if (unlikely(cork->opt == NULL))
 				return -ENOBUFS;
 		}
-		memcpy(cork->opt, opt, sizeof(struct ip_options) + opt->optlen);
+		memcpy(cork->opt, &opt->opt, sizeof(struct ip_options) + opt->opt.optlen);
 		cork->flags |= IPCORK_OPT;
 		cork->addr = ipc->addr;
 	}
@@ -1451,26 +1452,23 @@ void ip_send_reply(struct sock *sk, struct sk_buff *skb, struct ip_reply_arg *ar
 		   unsigned int len)
 {
 	struct inet_sock *inet = inet_sk(sk);
-	struct {
-		struct ip_options	opt;
-		char			data[40];
-	} replyopts;
+	struct ip_options_data replyopts;
 	struct ipcm_cookie ipc;
 	__be32 daddr;
 	struct rtable *rt = skb_rtable(skb);
 
-	if (ip_options_echo(&replyopts.opt, skb))
+	if (ip_options_echo(&replyopts.opt.opt, skb))
 		return;
 
 	daddr = ipc.addr = rt->rt_src;
 	ipc.opt = NULL;
 	ipc.tx_flags = 0;
 
-	if (replyopts.opt.optlen) {
+	if (replyopts.opt.opt.optlen) {
 		ipc.opt = &replyopts.opt;
 
-		if (ipc.opt->srr)
-			daddr = replyopts.opt.faddr;
+		if (replyopts.opt.opt.srr)
+			daddr = replyopts.opt.opt.faddr;
 	}
 
 	{
diff --git a/net/ipv4/ip_sockglue.c b/net/ipv4/ip_sockglue.c
index 9640900..ab0c9ef 100644
--- a/net/ipv4/ip_sockglue.c
+++ b/net/ipv4/ip_sockglue.c
@@ -451,6 +451,11 @@ out:
 }
 
 
+static void opt_kfree_rcu(struct rcu_head *head)
+{
+	kfree(container_of(head, struct ip_options_rcu, rcu));
+}
+
 /*
  *	Socket option code for IP. This is the end of the line after any
  *	TCP,UDP etc options on an IP socket.
@@ -497,13 +502,16 @@ static int do_ip_setsockopt(struct sock *sk, int level,
 	switch (optname) {
 	case IP_OPTIONS:
 	{
-		struct ip_options *opt = NULL;
+		struct ip_options_rcu *old, *opt = NULL;
+
 		if (optlen > 40)
 			goto e_inval;
 		err = ip_options_get_from_user(sock_net(sk), &opt,
 					       optval, optlen);
 		if (err)
 			break;
+		old = rcu_dereference_protected(inet->inet_opt,
+						sock_owned_by_user(sk));
 		if (inet->is_icsk) {
 			struct inet_connection_sock *icsk = inet_csk(sk);
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
@@ -512,17 +520,18 @@ static int do_ip_setsockopt(struct sock *sk, int level,
 			       (TCPF_LISTEN | TCPF_CLOSE)) &&
 			     inet->inet_daddr != LOOPBACK4_IPV6)) {
 #endif
-				if (inet->opt)
-					icsk->icsk_ext_hdr_len -= inet->opt->optlen;
+				if (old)
+					icsk->icsk_ext_hdr_len -= old->opt.optlen;
 				if (opt)
-					icsk->icsk_ext_hdr_len += opt->optlen;
+					icsk->icsk_ext_hdr_len += opt->opt.optlen;
 				icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 			}
 #endif
 		}
-		opt = xchg(&inet->opt, opt);
-		kfree(opt);
+		rcu_assign_pointer(inet->inet_opt, opt);
+		if (old)
+			call_rcu(&old->rcu, opt_kfree_rcu);
 		break;
 	}
 	case IP_PKTINFO:
@@ -1081,12 +1090,16 @@ static int do_ip_getsockopt(struct sock *sk, int level, int optname,
 	case IP_OPTIONS:
 	{
 		unsigned char optbuf[sizeof(struct ip_options)+40];
-		struct ip_options * opt = (struct ip_options *)optbuf;
+		struct ip_options *opt = (struct ip_options *)optbuf;
+		struct ip_options_rcu *inet_opt;
+
+		inet_opt = rcu_dereference_protected(inet->inet_opt,
+						     sock_owned_by_user(sk));
 		opt->optlen = 0;
-		if (inet->opt)
-			memcpy(optbuf, inet->opt,
-			       sizeof(struct ip_options)+
-			       inet->opt->optlen);
+		if (inet_opt)
+			memcpy(optbuf, &inet_opt->opt,
+			       sizeof(struct ip_options) +
+			       inet_opt->opt.optlen);
 		release_sock(sk);
 
 		if (opt->optlen == 0)
diff --git a/net/ipv4/raw.c b/net/ipv4/raw.c
index abf14db..a8659e0 100644
--- a/net/ipv4/raw.c
+++ b/net/ipv4/raw.c
@@ -460,6 +460,7 @@ static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 	__be32 saddr;
 	u8  tos;
 	int err;
+	struct ip_options_data opt_copy;
 
 	err = -EMSGSIZE;
 	if (len > 0xFFFF)
@@ -520,8 +521,18 @@ static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 	saddr = ipc.addr;
 	ipc.addr = daddr;
 
-	if (!ipc.opt)
-		ipc.opt = inet->opt;
+	if (!ipc.opt) {
+		struct ip_options_rcu *inet_opt;
+
+		rcu_read_lock();
+		inet_opt = rcu_dereference(inet->inet_opt);
+		if (inet_opt) {
+			memcpy(&opt_copy, inet_opt,
+			       sizeof(*inet_opt) + inet_opt->opt.optlen);
+			ipc.opt = &opt_copy.opt;
+		}
+		rcu_read_unlock();
+	}
 
 	if (ipc.opt) {
 		err = -EINVAL;
@@ -530,10 +541,10 @@ static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 		 */
 		if (inet->hdrincl)
 			goto done;
-		if (ipc.opt->srr) {
+		if (ipc.opt->opt.srr) {
 			if (!daddr)
 				goto done;
-			daddr = ipc.opt->faddr;
+			daddr = ipc.opt->opt.faddr;
 		}
 	}
 	tos = RT_CONN_FLAGS(sk);
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 71e0296..2646149 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -321,10 +321,10 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb,
 	 * the ACK carries the same options again (see RFC1122 4.2.3.8)
 	 */
 	if (opt && opt->optlen) {
-		int opt_size = sizeof(struct ip_options) + opt->optlen;
+		int opt_size = sizeof(struct ip_options_rcu) + opt->optlen;
 
 		ireq->opt = kmalloc(opt_size, GFP_ATOMIC);
-		if (ireq->opt != NULL && ip_options_echo(ireq->opt, skb)) {
+		if (ireq->opt != NULL && ip_options_echo(&ireq->opt->opt, skb)) {
 			kfree(ireq->opt);
 			ireq->opt = NULL;
 		}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 310454c..d60732f 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -154,6 +154,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	struct flowi4 fl4;
 	struct rtable *rt;
 	int err;
+	struct ip_options_rcu *inet_opt;
 
 	if (addr_len < sizeof(struct sockaddr_in))
 		return -EINVAL;
@@ -162,10 +163,12 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 		return -EAFNOSUPPORT;
 
 	nexthop = daddr = usin->sin_addr.s_addr;
-	if (inet->opt && inet->opt->srr) {
+	inet_opt = rcu_dereference_protected(inet->inet_opt,
+					     sock_owned_by_user(sk));
+	if (inet_opt && inet_opt->opt.srr) {
 		if (!daddr)
 			return -EINVAL;
-		nexthop = inet->opt->faddr;
+		nexthop = inet_opt->opt.faddr;
 	}
 
 	orig_sport = inet->inet_sport;
@@ -186,7 +189,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 		return -ENETUNREACH;
 	}
 
-	if (!inet->opt || !inet->opt->srr)
+	if (!inet_opt || !inet_opt->opt.srr)
 		daddr = rt->rt_dst;
 
 	if (!inet->inet_saddr)
@@ -222,8 +225,8 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 	inet->inet_daddr = daddr;
 
 	inet_csk(sk)->icsk_ext_hdr_len = 0;
-	if (inet->opt)
-		inet_csk(sk)->icsk_ext_hdr_len = inet->opt->optlen;
+	if (inet_opt)
+		inet_csk(sk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
 
 	tp->rx_opt.mss_clamp = TCP_MSS_DEFAULT;
 
@@ -820,17 +823,18 @@ static void syn_flood_warning(const struct sk_buff *skb)
 /*
  * Save and compile IPv4 options into the request_sock if needed.
  */
-static struct ip_options *tcp_v4_save_options(struct sock *sk,
-					      struct sk_buff *skb)
+static struct ip_options_rcu *tcp_v4_save_options(struct sock *sk,
+						  struct sk_buff *skb)
 {
-	struct ip_options *opt = &(IPCB(skb)->opt);
-	struct ip_options *dopt = NULL;
+	const struct ip_options *opt = &(IPCB(skb)->opt);
+	struct ip_options_rcu *dopt = NULL;
 
 	if (opt && opt->optlen) {
-		int opt_size = optlength(opt);
+		int opt_size = sizeof(*dopt) + opt->optlen;
+
 		dopt = kmalloc(opt_size, GFP_ATOMIC);
 		if (dopt) {
-			if (ip_options_echo(dopt, skb)) {
+			if (ip_options_echo(&dopt->opt, skb)) {
 				kfree(dopt);
 				dopt = NULL;
 			}
@@ -1411,6 +1415,7 @@ struct sock *tcp_v4_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 #ifdef CONFIG_TCP_MD5SIG
 	struct tcp_md5sig_key *key;
 #endif
+	struct ip_options_rcu *inet_opt;
 
 	if (sk_acceptq_is_full(sk))
 		goto exit_overflow;
@@ -1431,13 +1436,14 @@ struct sock *tcp_v4_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
 	newinet->inet_daddr   = ireq->rmt_addr;
 	newinet->inet_rcv_saddr = ireq->loc_addr;
 	newinet->inet_saddr	      = ireq->loc_addr;
-	newinet->opt	      = ireq->opt;
+	inet_opt	      = ireq->opt;
+	rcu_assign_pointer(newinet->inet_opt, inet_opt);
 	ireq->opt	      = NULL;
 	newinet->mc_index     = inet_iif(skb);
 	newinet->mc_ttl	      = ip_hdr(skb)->ttl;
 	inet_csk(newsk)->icsk_ext_hdr_len = 0;
-	if (newinet->opt)
-		inet_csk(newsk)->icsk_ext_hdr_len = newinet->opt->optlen;
+	if (inet_opt)
+		inet_csk(newsk)->icsk_ext_hdr_len = inet_opt->opt.optlen;
 	newinet->inet_id = newtp->write_seq ^ jiffies;
 
 	tcp_mtup_init(newsk);
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index bc0dab2..544f435 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -804,6 +804,7 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 	int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
 	int (*getfrag)(void *, char *, int, int, int, struct sk_buff *);
 	struct sk_buff *skb;
+	struct ip_options_data opt_copy;
 
 	if (len > 0xFFFF)
 		return -EMSGSIZE;
@@ -877,22 +878,32 @@ int udp_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 			free = 1;
 		connected = 0;
 	}
-	if (!ipc.opt)
-		ipc.opt = inet->opt;
+	if (!ipc.opt) {
+		struct ip_options_rcu *inet_opt;
+
+		rcu_read_lock();
+		inet_opt = rcu_dereference(inet->inet_opt);
+		if (inet_opt) {
+			memcpy(&opt_copy, inet_opt,
+			       sizeof(*inet_opt) + inet_opt->opt.optlen);
+			ipc.opt = &opt_copy.opt;
+		}
+		rcu_read_unlock();
+	}
 
 	saddr = ipc.addr;
 	ipc.addr = faddr = daddr;
 
-	if (ipc.opt && ipc.opt->srr) {
+	if (ipc.opt && ipc.opt->opt.srr) {
 		if (!daddr)
 			return -EINVAL;
-		faddr = ipc.opt->faddr;
+		faddr = ipc.opt->opt.faddr;
 		connected = 0;
 	}
 	tos = RT_TOS(inet->tos);
 	if (sock_flag(sk, SOCK_LOCALROUTE) ||
 	    (msg->msg_flags & MSG_DONTROUTE) ||
-	    (ipc.opt && ipc.opt->is_strictroute)) {
+	    (ipc.opt && ipc.opt->opt.is_strictroute)) {
 		tos |= RTO_ONLINK;
 		connected = 0;
 	}