[IPV4]: Respect hoplimit route metric.

include/linux/ip.h

@@ -113,7 +113,7 @@ struct inet_opt {
 	__u16			dport;		/* Destination port */
 	__u16			num;		/* Local port */
 	__u32			saddr;		/* Sending source */
-	int			ttl;		/* TTL setting */
+	int			uc_ttl;		/* Unicast TTL */
 	int			tos;		/* TOS */
 	unsigned	   	cmsg_flags;
 	struct ip_options	*opt;

net/ipv4/af_inet.c

@@ -398,7 +398,7 @@ static int inet_create(struct socket *sock, int protocol)
 	sk->protocol	= protocol;
 	sk->backlog_rcv = sk->prot->backlog_rcv;
 
-	inet->ttl	= sysctl_ip_default_ttl;
+	inet->uc_ttl	= -1;
 	inet->mc_loop	= 1;
 	inet->mc_ttl	= 1;
 	inet->mc_index	= 0;

net/ipv4/icmp.c

@@ -185,8 +185,6 @@ struct icmp_err icmp_err_convert[] = {
 	},
 };
 
-extern int sysctl_ip_default_ttl;
-
 /* Control parameters for ECHO replies. */
 int sysctl_icmp_echo_ignore_all;
 int sysctl_icmp_echo_ignore_broadcasts;
@@ -384,7 +382,6 @@ static void icmp_reply(struct icmp_bxm *icmp_param, struct sk_buff *skb)
 	icmp_out_count(icmp_param->data.icmph.type);
 
 	inet->tos = skb->nh.iph->tos;
-	inet->ttl = sysctl_ip_default_ttl;
 	daddr = ipc.addr = rt->rt_src;
 	ipc.opt = NULL;
 	if (icmp_param->replyopts.optlen) {
@@ -539,7 +536,6 @@ void icmp_send(struct sk_buff *skb_in, int type, int code, u32 info)
 	icmp_param.offset = skb_in->nh.raw - skb_in->data;
 	icmp_out_count(icmp_param.data.icmph.type);
 	inet_sk(icmp_socket->sk)->tos = tos;
-	inet_sk(icmp_socket->sk)->ttl = sysctl_ip_default_ttl;
 	ipc.addr = iph->saddr;
 	ipc.opt = &icmp_param.replyopts;
 	if (icmp_param.replyopts.srr) {
@@ -1126,7 +1122,7 @@ void __init icmp_init(struct net_proto_family *ops)
 		per_cpu(__icmp_socket, i)->sk->allocation = GFP_ATOMIC;
 		per_cpu(__icmp_socket, i)->sk->sndbuf = SK_WMEM_MAX * 2;
 		inet = inet_sk(per_cpu(__icmp_socket, i)->sk);
-		inet->ttl = MAXTTL;
+		inet->uc_ttl = -1;
 		inet->pmtudisc = IP_PMTUDISC_DONT;
 
 		/* Unhash it so that IP input processing does not even

net/ipv4/ip_gre.c

@@ -860,7 +860,7 @@ static int ipgre_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 			iph->ttl = ((struct ipv6hdr*)old_iph)->hop_limit;
 #endif
 		else
-			iph->ttl = sysctl_ip_default_ttl;
+			iph->ttl = dst_metric(&rt->u.dst, RTAX_HOPLIMIT);
 	}
 
 	((u16*)(iph+1))[0] = tunnel->parms.o_flags;

net/ipv4/ip_output.c

@@ -112,6 +112,15 @@ static int ip_dev_loopback_xmit(struct sk_buff *newskb)
 	return 0;
 }
 
+static inline int ip_select_ttl(struct inet_opt *inet, struct dst_entry *dst)
+{
+	int ttl = inet->uc_ttl;
+
+	if (ttl < 0)
+		ttl = dst_metric(dst, RTAX_HOPLIMIT);
+	return ttl;
+}
+
 /* 
  *		Add an ip header to a skbuff and send it out.
  *
@@ -136,7 +145,7 @@ int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
 		iph->frag_off = htons(IP_DF);
 	else
 		iph->frag_off = 0;
-	iph->ttl      = inet->ttl;
+	iph->ttl      = ip_select_ttl(inet, &rt->u.dst);
 	iph->daddr    = rt->rt_dst;
 	iph->saddr    = rt->rt_src;
 	iph->protocol = sk->protocol;
@@ -341,7 +350,7 @@ int ip_queue_xmit(struct sk_buff *skb, int ipfragok)
 		iph->frag_off = htons(IP_DF);
 	else
 		iph->frag_off = 0;
-	iph->ttl      = inet->ttl;
+	iph->ttl      = ip_select_ttl(inet, &rt->u.dst);
 	iph->protocol = sk->protocol;
 	iph->saddr    = rt->rt_src;
 	iph->daddr    = rt->rt_dst;
@@ -1120,7 +1129,7 @@ int ip_push_pending_frames(struct sock *sk)
 	if (rt->rt_type == RTN_MULTICAST)
 		ttl = inet->mc_ttl;
 	else
-		ttl = inet->ttl;
+		ttl = ip_select_ttl(inet, &rt->u.dst);
 
 	iph = (struct iphdr *)skb->data;
 	iph->version = 4;

net/ipv4/ip_sockglue.c

@@ -501,11 +501,9 @@ int ip_setsockopt(struct sock *sk, int level, int optname, char *optval, int opt
 		case IP_TTL:
 			if (optlen<1)
 				goto e_inval;
-			if(val==-1)
-				val = sysctl_ip_default_ttl;
-			if(val<1||val>255)
+			if (val != -1 && (val < 1 || val>255))
 				goto e_inval;
-			inet->ttl = val;
+			inet->uc_ttl = val;
 			break;
 		case IP_HDRINCL:
 			if(sk->type!=SOCK_RAW) {
@@ -911,7 +909,9 @@ int ip_getsockopt(struct sock *sk, int level, int optname, char *optval, int *op
 			val = inet->tos;
 			break;
 		case IP_TTL:
-			val = inet->ttl;
+			val = (inet->uc_ttl == -1 ?
+			       sysctl_ip_default_ttl :
+			       inet->uc_ttl);
 			break;
 		case IP_HDRINCL:
 			val = inet->hdrincl;

net/ipv4/proc.c

@@ -111,7 +111,6 @@ fold_field(void *mib[], int nr)
  */
 static int snmp_seq_show(struct seq_file *seq, void *v)
 {
-	extern int sysctl_ip_default_ttl;
 	int i;
 
 	seq_printf(seq, "Ip: Forwarding DefaultTTL InReceives InHdrErrors "

net/ipv4/route.c

@@ -1317,6 +1317,9 @@ static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
 			rt->rt_gateway = FIB_RES_GW(*res);
 		memcpy(rt->u.dst.metrics, fi->fib_metrics,
 		       sizeof(rt->u.dst.metrics));
+		if (rt->u.dst.metrics[RTAX_HOPLIMIT-1] == 0)
+			rt->u.dst.metrics[RTAX_HOPLIMIT-1] =
+				sysctl_ip_default_ttl;
 		if (fi->fib_mtu == 0) {
 			rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu;
 			if (rt->u.dst.metrics[RTAX_LOCK-1] & (1 << RTAX_MTU) &&

net/ipv4/tcp_ipv4.c

@@ -74,7 +74,6 @@
 #include <linux/seq_file.h>
 
 extern int sysctl_ip_dynaddr;
-extern int sysctl_ip_default_ttl;
 int sysctl_tcp_tw_reuse;
 int sysctl_tcp_low_latency;
 
@@ -1213,7 +1212,6 @@ static void tcp_v4_send_reset(struct sk_buff *skb)
 				      sizeof(struct tcphdr), IPPROTO_TCP, 0);
 	arg.csumoffset = offsetof(struct tcphdr, check) / 2;
 
-	inet_sk(tcp_socket->sk)->ttl = sysctl_ip_default_ttl;
 	ip_send_reply(tcp_socket->sk, skb, &arg, sizeof rth);
 
 	TCP_INC_STATS_BH(TcpOutSegs);
@@ -2619,7 +2617,7 @@ void __init tcp_v4_init(struct net_proto_family *ops)
 	if (err < 0)
 		panic("Failed to create the TCP control socket.\n");
 	tcp_socket->sk->allocation   = GFP_ATOMIC;
-	inet_sk(tcp_socket->sk)->ttl = MAXTTL;
+	inet_sk(tcp_socket->sk)->uc_ttl = -1;
 
 	/* Unhash it so that IP input processing does not even
 	 * see it, we do not wish this socket to see incoming

net/ipv6/af_inet6.c

@@ -227,7 +227,7 @@ static int inet6_create(struct socket *sock, int protocol)
 	/* Init the ipv4 part of the socket since we can have sockets
 	 * using v6 API for ipv4.
 	 */
-	inet->ttl	= 64;
+	inet->uc_ttl	= -1;
 
 	inet->mc_loop	= 1;
 	inet->mc_ttl	= 1;

net/sctp/ipv6.c

@@ -547,7 +547,7 @@ struct sock *sctp_v6_create_accept_sk(struct sock *sk,
 	/* Init the ipv4 part of the socket since we can have sockets
 	 * using v6 API for ipv4.
 	 */
-	newinet->ttl = sysctl_ip_default_ttl;
+	newinet->uc_ttl = -1;
 	newinet->mc_loop = 1;
 	newinet->mc_ttl = 1;
 	newinet->mc_index = 0;

net/sctp/protocol.c

@@ -545,7 +545,7 @@ struct sock *sctp_v4_create_accept_sk(struct sock *sk,
 	newinet->pmtudisc = inet->pmtudisc;
       	newinet->id = 0;
 
-	newinet->ttl = sysctl_ip_default_ttl;
+	newinet->uc_ttl = -1;
 	newinet->mc_loop = 1;
 	newinet->mc_ttl = 1;
 	newinet->mc_index = 0;
@@ -602,7 +602,7 @@ int sctp_ctl_sock_init(void)
 		return err;
 	}
 	sctp_ctl_socket->sk->allocation = GFP_ATOMIC;
-	inet_sk(sctp_ctl_socket->sk)->ttl = MAXTTL;
+	inet_sk(sctp_ctl_socket->sk)->uc_ttl = -1;
 
 	return 0;
 }