Merge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

11 files changed, 282 insertions, 142 deletions

diff --git a/net/Kconfig b/net/Kconfig
index 2bdd5623fdd..60f6f321bd7 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -140,6 +140,7 @@ config BRIDGE_NETFILTER
 
 	  If unsure, say N.
 
+source "net/netfilter/Kconfig"
 source "net/ipv4/netfilter/Kconfig"
 source "net/ipv6/netfilter/Kconfig"
 source "net/decnet/netfilter/Kconfig"
@@ -206,8 +207,6 @@ config NET_PKTGEN
 	  To compile this code as a module, choose M here: the
 	  module will be called pktgen.
 
-source "net/netfilter/Kconfig"
-
 endmenu
 
 endmenu
diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h
index 95c4630b3b1..be7a660b6b2 100644
--- a/net/dccp/dccp.h
+++ b/net/dccp/dccp.h
@@ -258,13 +258,12 @@ extern int	   dccp_v4_send_reset(struct sock *sk,
 extern void	   dccp_send_close(struct sock *sk, const int active);
 
 struct dccp_skb_cb {
-	__u8 dccpd_type;
-	__u8 dccpd_reset_code;
-	__u8 dccpd_service;
-	__u8 dccpd_ccval;
+	__u8  dccpd_type:4;
+	__u8  dccpd_ccval:4;
+	__u8  dccpd_reset_code;
+	__u16 dccpd_opt_len;
 	__u64 dccpd_seq;
 	__u64 dccpd_ack_seq;
-	int  dccpd_opt_len;
 };
 
 #define DCCP_SKB_CB(__skb) ((struct dccp_skb_cb *)&((__skb)->cb[0]))
diff --git a/net/dccp/input.c b/net/dccp/input.c
index c74034cf7ed..062e9f8359d 100644
--- a/net/dccp/input.c
+++ b/net/dccp/input.c
@@ -384,9 +384,9 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk,
 	}
 
 out_invalid_packet:
-	return 1; /* dccp_v4_do_rcv will send a reset, but...
-		     FIXME: the reset code should be
-			    DCCP_RESET_CODE_PACKET_ERROR */
+	/* dccp_v4_do_rcv will send a reset */
+	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;
+	return 1; 
 }
 
 static int dccp_rcv_respond_partopen_state_process(struct sock *sk,
@@ -433,6 +433,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 			   struct dccp_hdr *dh, unsigned len)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
 	const int old_state = sk->sk_state;
 	int queued = 0;
 
@@ -473,7 +474,8 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		if (dh->dccph_type == DCCP_PKT_RESET)
 			goto discard;
 
-		/* Caller (dccp_v4_do_rcv) will send Reset(No Connection)*/
+		/* Caller (dccp_v4_do_rcv) will send Reset */
+		dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
 		return 1;
 	}
 
@@ -487,8 +489,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		if (dccp_parse_options(sk, skb))
 			goto discard;
 
-		if (DCCP_SKB_CB(skb)->dccpd_ack_seq !=
-		    DCCP_PKT_WITHOUT_ACK_SEQ)
+		if (dcb->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
 			dccp_event_ack_recv(sk, skb);
 
 		ccid_hc_rx_packet_recv(dp->dccps_hc_rx_ccid, sk, skb);
@@ -500,7 +501,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		 */
 		if (dp->dccps_options.dccpo_send_ack_vector) {
 			if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk,
-					     DCCP_SKB_CB(skb)->dccpd_seq,
+					     dcb->dccpd_seq,
 					     DCCP_ACKPKTS_STATE_RECEIVED))
 				goto discard;
 			/*
@@ -551,8 +552,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 		     dh->dccph_type == DCCP_PKT_REQUEST) ||
 		    (sk->sk_state == DCCP_RESPOND &&
 		     dh->dccph_type == DCCP_PKT_DATA)) {
-		dccp_send_sync(sk, DCCP_SKB_CB(skb)->dccpd_seq,
-			       DCCP_PKT_SYNC);
+		dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNC);
 		goto discard;
 	} else if (dh->dccph_type == DCCP_PKT_CLOSEREQ) {
 		dccp_rcv_closereq(sk, skb);
@@ -563,13 +563,13 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
 	}
 
 	if (unlikely(dh->dccph_type == DCCP_PKT_SYNC)) {
-		dccp_send_sync(sk, DCCP_SKB_CB(skb)->dccpd_seq,
-			       DCCP_PKT_SYNCACK);
+		dccp_send_sync(sk, dcb->dccpd_seq, DCCP_PKT_SYNCACK);
 		goto discard;
 	}
 
 	switch (sk->sk_state) {
 	case DCCP_CLOSED:
+		dcb->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
 		return 1;
 
 	case DCCP_REQUESTING:
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index 2afaa464e7f..94a440b2685 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -246,6 +246,9 @@ static int dccp_v4_connect(struct sock *sk, struct sockaddr *uaddr,
 
 	dp->dccps_role = DCCP_ROLE_CLIENT;
 
+	if (dccp_service_not_initialized(sk))
+		return -EPROTO;
+
 	if (addr_len < sizeof(struct sockaddr_in))
 		return -EINVAL;
 
@@ -661,6 +664,16 @@ static inline u64 dccp_v4_init_sequence(const struct sock *sk,
 					   dccp_hdr(skb)->dccph_sport);
 }
 
+static inline int dccp_bad_service_code(const struct sock *sk,
+					const __u32 service)
+{
+	const struct dccp_sock *dp = dccp_sk(sk);
+
+	if (dp->dccps_service == service)
+		return 0;
+	return !dccp_list_has_service(dp->dccps_service_list, service);
+}
+
 int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 {
 	struct inet_request_sock *ireq;
@@ -669,13 +682,22 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	struct dccp_request_sock *dreq;
 	const __u32 saddr = skb->nh.iph->saddr;
 	const __u32 daddr = skb->nh.iph->daddr;
+ 	const __u32 service = dccp_hdr_request(skb)->dccph_req_service;
+	struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
+	__u8 reset_code = DCCP_RESET_CODE_TOO_BUSY;
 	struct dst_entry *dst = NULL;
 
 	/* Never answer to DCCP_PKT_REQUESTs send to broadcast or multicast */
 	if (((struct rtable *)skb->dst)->rt_flags &
-	    (RTCF_BROADCAST | RTCF_MULTICAST))
+	    (RTCF_BROADCAST | RTCF_MULTICAST)) {
+		reset_code = DCCP_RESET_CODE_NO_CONNECTION;
 		goto drop;
+	}
 
+	if (dccp_bad_service_code(sk, service)) {
+		reset_code = DCCP_RESET_CODE_BAD_SERVICE_CODE;
+		goto drop;
+ 	}
 	/*
 	 * TW buckets are converted to open requests without
 	 * limitations, they conserve resources and peer is
@@ -718,9 +740,9 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	 * dccp_create_openreq_child.
 	 */
 	dreq = dccp_rsk(req);
-	dreq->dreq_isr = DCCP_SKB_CB(skb)->dccpd_seq;
-	dreq->dreq_iss = dccp_v4_init_sequence(sk, skb);
-	dreq->dreq_service = dccp_hdr_request(skb)->dccph_req_service;
+	dreq->dreq_isr	   = dcb->dccpd_seq;
+	dreq->dreq_iss	   = dccp_v4_init_sequence(sk, skb);
+	dreq->dreq_service = service;
 
 	if (dccp_v4_send_response(sk, req, dst))
 		goto drop_and_free;
@@ -735,6 +757,7 @@ drop_and_free:
 	__reqsk_free(req);
 drop:
 	DCCP_INC_STATS_BH(DCCP_MIB_ATTEMPTFAILS);
+	dcb->dccpd_reset_code = reset_code;
 	return -1;
 }
 
@@ -1005,7 +1028,6 @@ int dccp_v4_do_rcv(struct sock *sk, struct sk_buff *skb)
 	return 0;
 
 reset:
-	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_NO_CONNECTION;
 	dccp_v4_ctl_send_reset(skb);
 discard:
 	kfree_skb(skb);
@@ -1280,6 +1302,7 @@ static int dccp_v4_init_sock(struct sock *sk)
 	sk->sk_write_space = dccp_write_space;
 	dp->dccps_mss_cache = 536;
 	dp->dccps_role = DCCP_ROLE_UNDEFINED;
+	dp->dccps_service = DCCP_SERVICE_INVALID_VALUE;
 
 	return 0;
 }
@@ -1301,6 +1324,11 @@ static int dccp_v4_destroy_sock(struct sock *sk)
 	if (inet_csk(sk)->icsk_bind_hash != NULL)
 		inet_put_port(&dccp_hashinfo, sk);
 
+	if (dp->dccps_service_list != NULL) {
+		kfree(dp->dccps_service_list);
+		dp->dccps_service_list = NULL;
+	}
+
 	ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk);
 	ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk);
 	dccp_ackpkts_free(dp->dccps_hc_rx_ackpkts);
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index 18461bc04cb..933e10db178 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -93,9 +93,11 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 		struct inet_connection_sock *newicsk = inet_csk(sk);
 		struct dccp_sock *newdp = dccp_sk(newsk);
 
+		newdp->dccps_role	   = DCCP_ROLE_SERVER;
 		newdp->dccps_hc_rx_ackpkts = NULL;
-		newdp->dccps_role = DCCP_ROLE_SERVER;
-		newicsk->icsk_rto = DCCP_TIMEOUT_INIT;
+		newdp->dccps_service_list  = NULL;
+		newdp->dccps_service	   = dreq->dreq_service;
+		newicsk->icsk_rto	   = DCCP_TIMEOUT_INIT;
 		do_gettimeofday(&newdp->dccps_epoch);
 
 		if (newdp->dccps_options.dccpo_send_ack_vector) {
diff --git a/net/dccp/output.c b/net/dccp/output.c
index ea6d0e91e51..156b1d29a15 100644
--- a/net/dccp/output.c
+++ b/net/dccp/output.c
@@ -85,7 +85,7 @@ int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
 		switch (dcb->dccpd_type) {
 		case DCCP_PKT_REQUEST:
 			dccp_hdr_request(skb)->dccph_req_service =
-							dcb->dccpd_service;
+							dp->dccps_service;
 			break;
 		case DCCP_PKT_RESET:
 			dccp_hdr_reset(skb)->dccph_reset_code =
@@ -270,6 +270,7 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
 				   struct request_sock *req)
 {
 	struct dccp_hdr *dh;
+	struct dccp_request_sock *dreq;
 	const int dccp_header_size = sizeof(struct dccp_hdr) +
 				     sizeof(struct dccp_hdr_ext) +
 				     sizeof(struct dccp_hdr_response);
@@ -285,8 +286,9 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
 	skb->dst = dst_clone(dst);
 	skb->csum = 0;
 
+	dreq = dccp_rsk(req);
 	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_RESPONSE;
-	DCCP_SKB_CB(skb)->dccpd_seq  = dccp_rsk(req)->dreq_iss;
+	DCCP_SKB_CB(skb)->dccpd_seq  = dreq->dreq_iss;
 	dccp_insert_options(sk, skb);
 
 	skb->h.raw = skb_push(skb, dccp_header_size);
@@ -300,8 +302,9 @@ struct sk_buff *dccp_make_response(struct sock *sk, struct dst_entry *dst,
 			   DCCP_SKB_CB(skb)->dccpd_opt_len) / 4;
 	dh->dccph_type	= DCCP_PKT_RESPONSE;
 	dh->dccph_x	= 1;
-	dccp_hdr_set_seq(dh, dccp_rsk(req)->dreq_iss);
-	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dccp_rsk(req)->dreq_isr);
+	dccp_hdr_set_seq(dh, dreq->dreq_iss);
+	dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), dreq->dreq_isr);
+	dccp_hdr_response(skb)->dccph_resp_service = dreq->dreq_service;
 
 	dh->dccph_checksum = dccp_v4_checksum(skb, inet_rsk(req)->loc_addr,
 					      inet_rsk(req)->rmt_addr);
@@ -397,9 +400,6 @@ int dccp_connect(struct sock *sk)
 	skb_reserve(skb, MAX_DCCP_HEADER);
 
 	DCCP_SKB_CB(skb)->dccpd_type = DCCP_PKT_REQUEST;
-	/* FIXME: set service to something meaningful, coming
-	 * from userspace*/
-	DCCP_SKB_CB(skb)->dccpd_service = 0;
 	skb->csum = 0;
 	skb_set_owner_w(skb, sk);
 
diff --git a/net/dccp/proto.c b/net/dccp/proto.c
index 18a0e69c9dc..9bda2868eba 100644
--- a/net/dccp/proto.c
+++ b/net/dccp/proto.c
@@ -94,7 +94,15 @@ EXPORT_SYMBOL_GPL(dccp_state_name);
 
 static inline int dccp_listen_start(struct sock *sk)
 {
-	dccp_sk(sk)->dccps_role = DCCP_ROLE_LISTEN;
+	struct dccp_sock *dp = dccp_sk(sk);
+
+	dp->dccps_role = DCCP_ROLE_LISTEN;
+	/*
+	 * Apps need to use setsockopt(DCCP_SOCKOPT_SERVICE)
+	 * before calling listen()
+	 */
+	if (dccp_service_not_initialized(sk))
+		return -EPROTO;
 	return inet_csk_listen_start(sk, TCP_SYNQ_HSIZE);
 }
 
@@ -202,6 +210,42 @@ int dccp_ioctl(struct sock *sk, int cmd, unsigned long arg)
 	return -ENOIOCTLCMD;
 }
 
+static int dccp_setsockopt_service(struct sock *sk, const u32 service,
+				   char __user *optval, int optlen)
+{
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_service_list *sl = NULL;
+
+	if (service == DCCP_SERVICE_INVALID_VALUE || 
+	    optlen > DCCP_SERVICE_LIST_MAX_LEN * sizeof(u32))
+		return -EINVAL;
+
+	if (optlen > sizeof(service)) {
+		sl = kmalloc(optlen, GFP_KERNEL);
+		if (sl == NULL)
+			return -ENOMEM;
+
+		sl->dccpsl_nr = optlen / sizeof(u32) - 1;
+		if (copy_from_user(sl->dccpsl_list,
+				   optval + sizeof(service),
+				   optlen - sizeof(service)) ||
+		    dccp_list_has_service(sl, DCCP_SERVICE_INVALID_VALUE)) {
+			kfree(sl);
+			return -EFAULT;
+		}
+	}
+
+	lock_sock(sk);
+	dp->dccps_service = service;
+
+	if (dp->dccps_service_list != NULL)
+		kfree(dp->dccps_service_list);
+
+	dp->dccps_service_list = sl;
+	release_sock(sk);
+	return 0;
+}
+
 int dccp_setsockopt(struct sock *sk, int level, int optname,
 		    char __user *optval, int optlen)
 {
@@ -218,8 +262,10 @@ int dccp_setsockopt(struct sock *sk, int level, int optname,
 	if (get_user(val, (int __user *)optval))
 		return -EFAULT;
 
-	lock_sock(sk);
+	if (optname == DCCP_SOCKOPT_SERVICE)
+		return dccp_setsockopt_service(sk, val, optval, optlen);
 
+	lock_sock(sk);
 	dp = dccp_sk(sk);
 	err = 0;
 
@@ -236,6 +282,37 @@ int dccp_setsockopt(struct sock *sk, int level, int optname,
 	return err;
 }
 
+static int dccp_getsockopt_service(struct sock *sk, int len,
+				   u32 __user *optval,
+				   int __user *optlen)
+{
+	const struct dccp_sock *dp = dccp_sk(sk);
+	const struct dccp_service_list *sl;
+	int err = -ENOENT, slen = 0, total_len = sizeof(u32);
+
+	lock_sock(sk);
+	if (dccp_service_not_initialized(sk))
+		goto out;
+
+	if ((sl = dp->dccps_service_list) != NULL) {
+		slen = sl->dccpsl_nr * sizeof(u32);
+		total_len += slen;
+	}
+
+	err = -EINVAL;
+	if (total_len > len)
+		goto out;
+
+	err = 0;
+	if (put_user(total_len, optlen) ||
+	    put_user(dp->dccps_service, optval) ||
+	    (sl != NULL && copy_to_user(optval + 1, sl->dccpsl_list, slen)))
+		err = -EFAULT;
+out:
+	release_sock(sk);
+	return err;
+}
+
 int dccp_getsockopt(struct sock *sk, int level, int optname,
 		    char __user *optval, int __user *optlen)
 {
@@ -248,6 +325,10 @@ int dccp_getsockopt(struct sock *sk, int level, int optname,
 	if (get_user(len, optlen))
 		return -EFAULT;
 
+	if (optname == DCCP_SOCKOPT_SERVICE)
+		return dccp_getsockopt_service(sk, len,
+					       (u32 __user *)optval, optlen);
+
 	len = min_t(unsigned int, len, sizeof(int));
 	if (len < 0)
 		return -EINVAL;
diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig
index 30aa8e2ee21..85190de5710 100644
--- a/net/ipv4/netfilter/Kconfig
+++ b/net/ipv4/netfilter/Kconfig
@@ -8,6 +8,7 @@ menu "IP: Netfilter Configuration"
 # connection tracking, helpers and protocols
 config IP_NF_CONNTRACK
 	tristate "Connection tracking (required for masq/NAT)"
+	select NETFILTER_NETLINK if IP_NF_CONNTRACK_NETLINK!=n
 	---help---
 	  Connection tracking keeps a record of what packets have passed
 	  through your machine, in order to figure out how they are related
@@ -51,6 +52,15 @@ config IP_NF_CONNTRACK_EVENTS
 	  
 	  IF unsure, say `N'.
 
+config IP_NF_CONNTRACK_NETLINK
+	tristate 'Connection tracking netlink interface'
+	depends on IP_NF_CONNTRACK && NETFILTER_NETLINK
+	default IP_NF_CONNTRACK if NETFILTER_NETLINK=y
+	default m if NETFILTER_NETLINK=m
+	help
+	  This option enables support for a netlink-based userspace interface
+
+
 config IP_NF_CT_PROTO_SCTP
 	tristate  'SCTP protocol connection tracking support (EXPERIMENTAL)'
 	depends on IP_NF_CONNTRACK && EXPERIMENTAL
@@ -774,11 +784,5 @@ config IP_NF_ARP_MANGLE
 	  Allows altering the ARP packet payload: source and destination
 	  hardware and network addresses.
 
-config IP_NF_CONNTRACK_NETLINK
-        tristate 'Connection tracking netlink interface'
-        depends on IP_NF_CONNTRACK && NETFILTER_NETLINK
-        help
-          This option enables support for a netlink-based userspace interface
-
 endmenu
 
diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c
index 19cba16e6e1..f8cd8e42961 100644
--- a/net/ipv4/netfilter/ip_conntrack_core.c
+++ b/net/ipv4/netfilter/ip_conntrack_core.c
@@ -1143,7 +1143,10 @@ void ip_ct_refresh_acct(struct ip_conntrack *ct,
 		if (del_timer(&ct->timeout)) {
 			ct->timeout.expires = jiffies + extra_jiffies;
 			add_timer(&ct->timeout);
-			ip_conntrack_event_cache(IPCT_REFRESH, skb);
+			/* FIXME: We loose some REFRESH events if this function
+			 * is called without an skb.  I'll fix this later -HW */
+			if (skb)
+				ip_conntrack_event_cache(IPCT_REFRESH, skb);
 		}
 		ct_add_counters(ct, ctinfo, skb);
 		write_unlock_bh(&ip_conntrack_lock);
diff --git a/net/ipv4/netfilter/ipt_CLUSTERIP.c b/net/ipv4/netfilter/ipt_CLUSTERIP.c
index 7d38913754b..9bcb398fbc1 100644
--- a/net/ipv4/netfilter/ipt_CLUSTERIP.c
+++ b/net/ipv4/netfilter/ipt_CLUSTERIP.c
@@ -13,6 +13,7 @@
 #include <linux/config.h>
 #include <linux/proc_fs.h>
 #include <linux/jhash.h>
+#include <linux/bitops.h>
 #include <linux/skbuff.h>
 #include <linux/ip.h>
 #include <linux/tcp.h>
@@ -30,7 +31,7 @@
 #include <linux/netfilter_ipv4/ipt_CLUSTERIP.h>
 #include <linux/netfilter_ipv4/ip_conntrack.h>
 
-#define CLUSTERIP_VERSION "0.7"
+#define CLUSTERIP_VERSION "0.8"
 
 #define DEBUG_CLUSTERIP
 
@@ -49,13 +50,14 @@ MODULE_DESCRIPTION("iptables target for CLUSTERIP");
 struct clusterip_config {
 	struct list_head list;			/* list of all configs */
 	atomic_t refcount;			/* reference count */
+	atomic_t entries;			/* number of entries/rules
+						 * referencing us */
 
 	u_int32_t clusterip;			/* the IP address */
 	u_int8_t clustermac[ETH_ALEN];		/* the MAC address */
 	struct net_device *dev;			/* device */
 	u_int16_t num_total_nodes;		/* total number of nodes */
-	u_int16_t num_local_nodes;		/* number of local nodes */
-	u_int16_t local_nodes[CLUSTERIP_MAX_NODES];	/* node number array */
+	unsigned long local_nodes;		/* node number array */
 
 #ifdef CONFIG_PROC_FS
 	struct proc_dir_entry *pde;		/* proc dir entry */
@@ -66,8 +68,7 @@ struct clusterip_config {
 
 static LIST_HEAD(clusterip_configs);
 
-/* clusterip_lock protects the clusterip_configs list _AND_ the configurable
- * data within all structurses (num_local_nodes, local_nodes[]) */
+/* clusterip_lock protects the clusterip_configs list */
 static DEFINE_RWLOCK(clusterip_lock);
 
 #ifdef CONFIG_PROC_FS
@@ -76,23 +77,48 @@ static struct proc_dir_entry *clusterip_procdir;
 #endif
 
 static inline void
-clusterip_config_get(struct clusterip_config *c) {
+clusterip_config_get(struct clusterip_config *c)
+{
 	atomic_inc(&c->refcount);
 }
 
 static inline void
-clusterip_config_put(struct clusterip_config *c) {
-	if (atomic_dec_and_test(&c->refcount)) {
+clusterip_config_put(struct clusterip_config *c)
+{
+	if (atomic_dec_and_test(&c->refcount))
+		kfree(c);
+}
+
+/* increase the count of entries(rules) using/referencing this config */
+static inline void
+clusterip_config_entry_get(struct clusterip_config *c)
+{
+	atomic_inc(&c->entries);
+}
+
+/* decrease the count of entries using/referencing this config.  If last
+ * entry(rule) is removed, remove the config from lists, but don't free it
+ * yet, since proc-files could still be holding references */
+static inline void
+clusterip_config_entry_put(struct clusterip_config *c)
+{
+	if (atomic_dec_and_test(&c->entries)) {
 		write_lock_bh(&clusterip_lock);
 		list_del(&c->list);
 		write_unlock_bh(&clusterip_lock);
+
 		dev_mc_delete(c->dev, c->clustermac, ETH_ALEN, 0);
 		dev_put(c->dev);
-		kfree(c);
+
+		/* In case anyone still accesses the file, the open/close
+		 * functions are also incrementing the refcount on their own,
+		 * so it's safe to remove the entry even if it's in use. */
+#ifdef CONFIG_PROC_FS
+		remove_proc_entry(c->pde->name, c->pde->parent);
+#endif
 	}
 }
 
-
 static struct clusterip_config *
 __clusterip_config_find(u_int32_t clusterip)
 {
@@ -111,7 +137,7 @@ __clusterip_config_find(u_int32_t clusterip)
 }
 
 static inline struct clusterip_config *
-clusterip_config_find_get(u_int32_t clusterip)
+clusterip_config_find_get(u_int32_t clusterip, int entry)
 {
 	struct clusterip_config *c;
 
@@ -122,11 +148,24 @@ clusterip_config_find_get(u_int32_t clusterip)
 		return NULL;
 	}
 	atomic_inc(&c->refcount);
+	if (entry)
+		atomic_inc(&c->entries);
 	read_unlock_bh(&clusterip_lock);
 
 	return c;
 }
 
+static void
+clusterip_config_init_nodelist(struct clusterip_config *c,
+			       const struct ipt_clusterip_tgt_info *i)
+{
+	int n;
+
+	for (n = 0; n < i->num_local_nodes; n++) {
+		set_bit(i->local_nodes[n] - 1, &c->local_nodes);
+	}
+}
+
 static struct clusterip_config *
 clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip,
 			struct net_device *dev)
@@ -143,11 +182,11 @@ clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip,
 	c->clusterip = ip;
 	memcpy(&c->clustermac, &i->clustermac, ETH_ALEN);
 	c->num_total_nodes = i->num_total_nodes;
-	c->num_local_nodes = i->num_local_nodes;
-	memcpy(&c->local_nodes, &i->local_nodes, sizeof(c->local_nodes));
+	clusterip_config_init_nodelist(c, i);
 	c->hash_mode = i->hash_mode;
 	c->hash_initval = i->hash_initval;
 	atomic_set(&c->refcount, 1);
+	atomic_set(&c->entries, 1);
 
 #ifdef CONFIG_PROC_FS
 	/* create proc dir entry */
@@ -171,53 +210,28 @@ clusterip_config_init(struct ipt_clusterip_tgt_info *i, u_int32_t ip,
 static int
 clusterip_add_node(struct clusterip_config *c, u_int16_t nodenum)
 {
-	int i;
-
-	write_lock_bh(&clusterip_lock);
 
-	if (c->num_local_nodes >= CLUSTERIP_MAX_NODES
-	    || nodenum > CLUSTERIP_MAX_NODES) {
-		write_unlock_bh(&clusterip_lock);
+	if (nodenum == 0 ||
+	    nodenum > c->num_total_nodes)
 		return 1;
-	}
-
-	/* check if we alrady have this number in our array */
-	for (i = 0; i < c->num_local_nodes; i++) {
-		if (c->local_nodes[i] == nodenum) {
-			write_unlock_bh(&clusterip_lock);
-			return 1;
-		}
-	}
 
-	c->local_nodes[c->num_local_nodes++] = nodenum;
+	/* check if we already have this number in our bitfield */
+	if (test_and_set_bit(nodenum - 1, &c->local_nodes))
+		return 1;
 
-	write_unlock_bh(&clusterip_lock);
 	return 0;
 }
 
 static int
 clusterip_del_node(struct clusterip_config *c, u_int16_t nodenum)
 {
-	int i;
-
-	write_lock_bh(&clusterip_lock);
-
-	if (c->num_local_nodes <= 1 || nodenum > CLUSTERIP_MAX_NODES) {
-		write_unlock_bh(&clusterip_lock);
+	if (nodenum == 0 ||
+	    nodenum > c->num_total_nodes)
 		return 1;
-	}
 		
-	for (i = 0; i < c->num_local_nodes; i++) {
-		if (c->local_nodes[i] == nodenum) {
-			int size = sizeof(u_int16_t)*(c->num_local_nodes-(i+1));
-			memmove(&c->local_nodes[i], &c->local_nodes[i+1], size);
-			c->num_local_nodes--;
-			write_unlock_bh(&clusterip_lock);
-			return 0;
-		}
-	}
+	if (test_and_clear_bit(nodenum - 1, &c->local_nodes))
+		return 0;
 
-	write_unlock_bh(&clusterip_lock);
 	return 1;
 }
 
@@ -285,25 +299,7 @@ clusterip_hashfn(struct sk_buff *skb, struct clusterip_config *config)
 static inline int
 clusterip_responsible(struct clusterip_config *config, u_int32_t hash)
 {
-	int i;
-
-	read_lock_bh(&clusterip_lock);
-
-	if (config->num_local_nodes == 0) {
-		read_unlock_bh(&clusterip_lock);
-		return 0;
-	}
-
-	for (i = 0; i < config->num_local_nodes; i++) {
-		if (config->local_nodes[i] == hash) {
-			read_unlock_bh(&clusterip_lock);
-			return 1;
-		}
-	}
-
-	read_unlock_bh(&clusterip_lock);
-
-	return 0;
+	return test_bit(hash - 1, &config->local_nodes);
 }
 
 /*********************************************************************** 
@@ -415,8 +411,26 @@ checkentry(const char *tablename,
 
 	/* FIXME: further sanity checks */
 
-	config = clusterip_config_find_get(e->ip.dst.s_addr);
-	if (!config) {
+	config = clusterip_config_find_get(e->ip.dst.s_addr, 1);
+	if (config) {
+		if (cipinfo->config != NULL) {
+			/* Case A: This is an entry that gets reloaded, since
+			 * it still has a cipinfo->config pointer. Simply
+			 * increase the entry refcount and return */
+			if (cipinfo->config != config) {
+				printk(KERN_ERR "CLUSTERIP: Reloaded entry "
+				       "has invalid config pointer!\n");
+				return 0;
+			}
+			clusterip_config_entry_get(cipinfo->config);
+		} else {
+			/* Case B: This is a new rule referring to an existing
+			 * clusterip config. */
+			cipinfo->config = config;
+			clusterip_config_entry_get(cipinfo->config);
+		}
+	} else {
+		/* Case C: This is a completely new clusterip config */
 		if (!(cipinfo->flags & CLUSTERIP_FLAG_NEW)) {
 			printk(KERN_WARNING "CLUSTERIP: no config found for %u.%u.%u.%u, need 'new'\n", NIPQUAD(e->ip.dst.s_addr));
 			return 0;
@@ -443,10 +457,9 @@ checkentry(const char *tablename,
 			}
 			dev_mc_add(config->dev,config->clustermac, ETH_ALEN, 0);
 		}
+		cipinfo->config = config;
 	}
 
-	cipinfo->config = config;
-
 	return 1;
 }
 
@@ -455,13 +468,10 @@ static void destroy(void *matchinfo, unsigned int matchinfosize)
 {
 	struct ipt_clusterip_tgt_info *cipinfo = matchinfo;
 
-	/* we first remove the proc entry and then drop the reference
-	 * count.  In case anyone still accesses the file, the open/close
-	 * functions are also incrementing the refcount on their own */
-#ifdef CONFIG_PROC_FS
-	remove_proc_entry(cipinfo->config->pde->name,
-			  cipinfo->config->pde->parent);
-#endif
+	/* if no more entries are referencing the config, remove it
+	 * from the list and destroy the proc entry */
+	clusterip_config_entry_put(cipinfo->config);
+
 	clusterip_config_put(cipinfo->config);
 }
 
@@ -533,7 +543,7 @@ arp_mangle(unsigned int hook,
 
 	/* if there is no clusterip configuration for the arp reply's 
 	 * source ip, we don't want to mangle it */
-	c = clusterip_config_find_get(payload->src_ip);
+	c = clusterip_config_find_get(payload->src_ip, 0);
 	if (!c)
 		return NF_ACCEPT;
 
@@ -574,56 +584,69 @@ static struct nf_hook_ops cip_arp_ops = {
 
 #ifdef CONFIG_PROC_FS
 
+struct clusterip_seq_position {
+	unsigned int pos;	/* position */
+	unsigned int weight;	/* number of bits set == size */
+	unsigned int bit;	/* current bit */
+	unsigned long val;	/* current value */
+};
+
 static void *clusterip_seq_start(struct seq_file *s, loff_t *pos)
 {
 	struct proc_dir_entry *pde = s->private;
 	struct clusterip_config *c = pde->data;
-	unsigned int *nodeidx;
-
-	read_lock_bh(&clusterip_lock);
-	if (*pos >= c->num_local_nodes)
+	unsigned int weight;
+	u_int32_t local_nodes;
+	struct clusterip_seq_position *idx;
+
+	/* FIXME: possible race */
+	local_nodes = c->local_nodes;
+	weight = hweight32(local_nodes);
+	if (*pos >= weight)
 		return NULL;
 
-	nodeidx = kmalloc(sizeof(unsigned int), GFP_KERNEL);
-	if (!nodeidx)
+	idx = kmalloc(sizeof(struct clusterip_seq_position), GFP_KERNEL);
+	if (!idx)
 		return ERR_PTR(-ENOMEM);
 
-	*nodeidx = *pos;
-	return nodeidx;
+	idx->pos = *pos;
+	idx->weight = weight;
+	idx->bit = ffs(local_nodes);
+	idx->val = local_nodes;
+	clear_bit(idx->bit - 1, &idx->val);
+
+	return idx;
 }
 
 static void *clusterip_seq_next(struct seq_file *s, void *v, loff_t *pos)
 {
-	struct proc_dir_entry *pde = s->private;
-	struct clusterip_config *c = pde->data;
-	unsigned int *nodeidx = (unsigned int *)v;
+	struct clusterip_seq_position *idx = (struct clusterip_seq_position *)v;
 
-	*pos = ++(*nodeidx);
-	if (*pos >= c->num_local_nodes) {
+	*pos = ++idx->pos;
+	if (*pos >= idx->weight) {
 		kfree(v);
 		return NULL;
 	}
-	return nodeidx;
+	idx->bit = ffs(idx->val);
+	clear_bit(idx->bit - 1, &idx->val);
+	return idx;
 }
 
 static void clusterip_seq_stop(struct seq_file *s, void *v)
 {
 	kfree(v);
-
-	read_unlock_bh(&clusterip_lock);
 }
 
 static int clusterip_seq_show(struct seq_file *s, void *v)
 {
-	struct proc_dir_entry *pde = s->private;
-	struct clusterip_config *c = pde->data;
-	unsigned int *nodeidx = (unsigned int *)v;
+	struct clusterip_seq_position *idx = (struct clusterip_seq_position *)v;
 
-	if (*nodeidx != 0) 
+	if (idx->pos != 0) 
 		seq_putc(s, ',');
-	seq_printf(s, "%u", c->local_nodes[*nodeidx]);
 
-	if (*nodeidx == c->num_local_nodes-1)
+	seq_printf(s, "%u", idx->bit);
+
+	if (idx->pos == idx->weight - 1)
 		seq_putc(s, '\n');
 
 	return 0;
diff --git a/net/socket.c b/net/socket.c
index c699e93c33d..f9264472377 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -1862,7 +1862,8 @@ asmlinkage long sys_recvmsg(int fd, struct msghdr __user *msg, unsigned int flag
 		if (err < 0)
 			goto out_freeiov;
 	}
-	err = __put_user(msg_sys.msg_flags, COMPAT_FLAGS(msg));
+	err = __put_user((msg_sys.msg_flags & ~MSG_CMSG_COMPAT),
+			 COMPAT_FLAGS(msg));
 	if (err)
 		goto out_freeiov;
 	if (MSG_CMSG_COMPAT & flags)