[DCCP]: Initial feature negotiation implementation

Still needs more work, but boots and doesn't crashes, even
does some negotiation!

18:38:52.174934  127.0.0.1.43458 > 127.0.0.1.5001: request <change_l ack_ratio 2, change_r ccid 2, change_l ccid 2>
18:38:52.218526  127.0.0.1.5001 > 127.0.0.1.43458: response <nop, nop, change_l ack_ratio 2, confirm_r ccid 2 2, confirm_l ccid 2 2, confirm_r ack_ratio 2>
18:38:52.185398  127.0.0.1.43458 > 127.0.0.1.5001: <nop, confirm_r ack_ratio 2, ack_vector0 0x00, elapsed_time 212>

:-)
Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/dccp.h

@@ -154,6 +154,10 @@ enum {
 	DCCPO_MANDATORY = 1,
 	DCCPO_MIN_RESERVED = 3,
 	DCCPO_MAX_RESERVED = 31,
+	DCCPO_CHANGE_L = 32,
+	DCCPO_CONFIRM_L = 33,
+	DCCPO_CHANGE_R = 34,
+	DCCPO_CONFIRM_R = 35,
 	DCCPO_NDP_COUNT = 37,
 	DCCPO_ACK_VECTOR_0 = 38,
 	DCCPO_ACK_VECTOR_1 = 39,
@@ -168,7 +172,9 @@ enum {
 /* DCCP features */
 enum {
 	DCCPF_RESERVED = 0,
+	DCCPF_CCID = 1,
 	DCCPF_SEQUENCE_WINDOW = 3,
+	DCCPF_ACK_RATIO = 5,
 	DCCPF_SEND_ACK_VECTOR = 6,
 	DCCPF_SEND_NDP_COUNT = 7,
 	/* 10-127 reserved */
@@ -176,9 +182,18 @@ enum {
 	DCCPF_MAX_CCID_SPECIFIC = 255,
 };
 
+/* this structure is argument to DCCP_SOCKOPT_CHANGE_X */
+struct dccp_so_feat {
+	__u8 dccpsf_feat;
+	__u8 *dccpsf_val;
+	__u8 dccpsf_len;
+};
+
 /* DCCP socket options */
 #define DCCP_SOCKOPT_PACKET_SIZE	1
 #define DCCP_SOCKOPT_SERVICE		2
+#define DCCP_SOCKOPT_CHANGE_L		3
+#define DCCP_SOCKOPT_CHANGE_R		4
 #define DCCP_SOCKOPT_CCID_RX_INFO	128
 #define DCCP_SOCKOPT_CCID_TX_INFO	192
 
@@ -314,8 +329,8 @@ static inline unsigned int dccp_hdr_len(const struct sk_buff *skb)
 
 /* initial values for each feature */
 #define DCCPF_INITIAL_SEQUENCE_WINDOW		100
-/* FIXME: for now we're using CCID 2 (TCP-Like) */
 #define DCCPF_INITIAL_CCID			2
+#define DCCPF_INITIAL_ACK_RATIO			2
 #define DCCPF_INITIAL_SEND_ACK_VECTOR		1
 /* FIXME: for now we're default to 1 but it should really be 0 */
 #define DCCPF_INITIAL_SEND_NDP_COUNT		1
@@ -335,6 +350,24 @@ struct dccp_options {
 	__u8	dccpo_tx_ccid;
 	__u8	dccpo_send_ack_vector;
 	__u8	dccpo_send_ndp_count;
+	__u8			dccpo_ack_ratio;
+	struct list_head	dccpo_pending;
+	struct list_head	dccpo_conf;
+};
+
+struct dccp_opt_conf {
+	__u8			*dccpoc_val;
+	__u8			dccpoc_len;
+};
+
+struct dccp_opt_pend {
+	struct list_head	dccpop_node;
+	__u8			dccpop_type;
+	__u8			dccpop_feat;
+	__u8		        *dccpop_val;
+	__u8			dccpop_len;
+	int			dccpop_conf;
+	struct dccp_opt_conf    *dccpop_sc;
 };
 
 extern void __dccp_options_init(struct dccp_options *dccpo);

net/dccp/Makefile

@@ -4,7 +4,7 @@ dccp_ipv6-y := ipv6.o
 
 obj-$(CONFIG_IP_DCCP) += dccp.o
 
-dccp-y := ccid.o input.o ipv4.o minisocks.o options.o output.o proto.o \
+dccp-y := ccid.o feat.o input.o ipv4.o minisocks.o options.o output.o proto.o \
 	  timer.o
 
 dccp-$(CONFIG_IP_DCCP_ACKVEC) += ackvec.o

net/dccp/feat.h

+#ifndef _DCCP_FEAT_H
+#define _DCCP_FEAT_H
+/*
+ *  net/dccp/feat.h
+ *
+ *  An implementation of the DCCP protocol
+ *  Copyright (c) 2005 Andrea Bittau <a.bittau@cs.ucl.ac.uk>
+ *
+ *	This program is free software; you can redistribute it and/or modify it
+ *	under the terms of the GNU General Public License version 2 as
+ *	published by the Free Software Foundation.
+ */
+
+#include <linux/types.h>
+
+struct sock;
+
+extern int  dccp_feat_change(struct sock *sk, u8 type, u8 feature,
+			     u8 *val, u8 len, gfp_t gfp);
+extern int  dccp_feat_change_recv(struct sock *sk, u8 type, u8 feature,
+				  u8 *val, u8 len);
+extern int  dccp_feat_confirm_recv(struct sock *sk, u8 type, u8 feature,
+				   u8 *val, u8 len);
+extern void dccp_feat_clean(struct sock *sk);
+extern int  dccp_feat_clone(struct sock *oldsk, struct sock *newsk);
+extern int  dccp_feat_init(struct sock *sk);
+
+#endif /* _DCCP_FEAT_H */

net/dccp/input.c

@@ -300,6 +300,9 @@ static int dccp_rcv_request_sent_state_process(struct sock *sk,
 			goto out_invalid_packet;
 		}
 
+		if (dccp_parse_options(sk, skb))
+			goto out_invalid_packet;
+
                 if (dp->dccps_options.dccpo_send_ack_vector &&
                     dccp_ackvec_add(dp->dccps_hc_rx_ackvec, sk,
                                     DCCP_SKB_CB(skb)->dccpd_seq,

net/dccp/ipv4.c

@@ -28,6 +28,7 @@
 #include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
+#include "feat.h"
 
 struct inet_hashinfo __cacheline_aligned dccp_hashinfo = {
 	.lhash_lock	= RW_LOCK_UNLOCKED,
@@ -535,7 +536,8 @@ int dccp_v4_conn_request(struct sock *sk, struct sk_buff *skb)
 	if (req == NULL)
 		goto drop;
 
-	/* FIXME: process options */
+	if (dccp_parse_options(sk, skb))
+		goto drop;
 
 	dccp_openreq_init(req, &dp, skb);
 
@@ -1049,6 +1051,8 @@ int dccp_v4_init_sock(struct sock *sk)
 	 * setsockopt(CCIDs-I-want/accept). -acme
 	 */
 	if (likely(!dccp_ctl_socket_init)) {
+		int rc;
+
 		if (dp->dccps_options.dccpo_send_ack_vector) {
 			dp->dccps_hc_rx_ackvec = dccp_ackvec_alloc(GFP_KERNEL);
 			if (dp->dccps_hc_rx_ackvec == NULL)
@@ -1069,8 +1073,16 @@ int dccp_v4_init_sock(struct sock *sk)
 			dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;
 			return -ENOMEM;
 		}
-	} else
+
+		rc = dccp_feat_init(sk);
+		if (rc)
+			return rc;
+	} else {
+		/* control socket doesn't need feat nego */
+		INIT_LIST_HEAD(&dp->dccps_options.dccpo_pending);
+		INIT_LIST_HEAD(&dp->dccps_options.dccpo_conf);
 		dccp_ctl_socket_init = 0;
+	}
 
 	dccp_init_xmit_timers(sk);
 	icsk->icsk_rto = DCCP_TIMEOUT_INIT;
@@ -1118,6 +1130,9 @@ int dccp_v4_destroy_sock(struct sock *sk)
 	ccid_exit(dp->dccps_hc_tx_ccid, sk);
 	dp->dccps_hc_rx_ccid = dp->dccps_hc_tx_ccid = NULL;
 
+	/* clean up feature negotiation state */
+	dccp_feat_clean(sk);
+
 	return 0;
 }
 

net/dccp/minisocks.c

@@ -22,6 +22,7 @@
 #include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
+#include "feat.h"
 
 struct inet_timewait_death_row dccp_death_row = {
 	.sysctl_max_tw_buckets = NR_FILE * 2,
@@ -114,6 +115,9 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 		newicsk->icsk_rto	   = DCCP_TIMEOUT_INIT;
 		do_gettimeofday(&newdp->dccps_epoch);
 
+		if (dccp_feat_clone(sk, newsk))
+			goto out_free;
+
 		if (newdp->dccps_options.dccpo_send_ack_vector) {
 			newdp->dccps_hc_rx_ackvec =
 						dccp_ackvec_alloc(GFP_ATOMIC);

net/dccp/options.c

@@ -21,12 +21,14 @@
 #include "ackvec.h"
 #include "ccid.h"
 #include "dccp.h"
+#include "feat.h"
 
 /* stores the default values for new connection. may be changed with sysctl */
 static const struct dccp_options dccpo_default_values = {
 	.dccpo_sequence_window	  = DCCPF_INITIAL_SEQUENCE_WINDOW,
 	.dccpo_rx_ccid		  = DCCPF_INITIAL_CCID,
 	.dccpo_tx_ccid		  = DCCPF_INITIAL_CCID,
+	.dccpo_ack_ratio	  = DCCPF_INITIAL_ACK_RATIO,
 	.dccpo_send_ack_vector	  = DCCPF_INITIAL_SEND_ACK_VECTOR,
 	.dccpo_send_ndp_count	  = DCCPF_INITIAL_SEND_NDP_COUNT,
 };
@@ -69,6 +71,8 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
 	unsigned char opt, len;
 	unsigned char *value;
 	u32 elapsed_time;
+	int rc;
+	int mandatory = 0;
 
 	memset(opt_recv, 0, sizeof(*opt_recv));
 
@@ -100,6 +104,11 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
 		switch (opt) {
 		case DCCPO_PADDING:
 			break;
+		case DCCPO_MANDATORY:
+			if (mandatory)
+				goto out_invalid_option;
+			mandatory = 1;
+			break;
 		case DCCPO_NDP_COUNT:
 			if (len > 3)
 				goto out_invalid_option;
@@ -108,6 +117,31 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
 			dccp_pr_debug("%sNDP count=%d\n", debug_prefix,
 				      opt_recv->dccpor_ndp);
 			break;
+		case DCCPO_CHANGE_L:
+			/* fall through */
+		case DCCPO_CHANGE_R:
+			if (len < 2)
+				goto out_invalid_option;
+			rc = dccp_feat_change_recv(sk, opt, *value, value + 1,
+						   len - 1);
+			/*
+			 * When there is a change error, change_recv is
+			 * responsible for dealing with it.  i.e. reply with an
+			 * empty confirm.
+			 * If the change was mandatory, then we need to die.
+			 */
+			if (rc && mandatory)
+				goto out_invalid_option;
+			break;
+		case DCCPO_CONFIRM_L:
+			/* fall through */
+		case DCCPO_CONFIRM_R:
+			if (len < 2)
+				goto out_invalid_option;
+			if (dccp_feat_confirm_recv(sk, opt, *value,
+						   value + 1, len - 1))
+				goto out_invalid_option;
+			break;
 		case DCCPO_ACK_VECTOR_0:
 		case DCCPO_ACK_VECTOR_1:
 			if (pkt_type == DCCP_PKT_DATA)
@@ -208,6 +242,9 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
 				sk, opt, len);
 			break;
 	        }
+
+		if (opt != DCCPO_MANDATORY)
+			mandatory = 0;
 	}
 
 	return 0;
@@ -356,7 +393,7 @@ void dccp_insert_option_timestamp(struct sock *sk, struct sk_buff *skb)
 {
 	struct timeval tv;
 	u32 now;
-	
+
 	dccp_timestamp(sk, &tv);
 	now = timeval_usecs(&tv) / 10;
 	/* yes this will overflow but that is the point as we want a
@@ -402,7 +439,7 @@ static void dccp_insert_option_timestamp_echo(struct sock *sk,
 	tstamp_echo = htonl(dp->dccps_timestamp_echo);
 	memcpy(to, &tstamp_echo, 4);
 	to += 4;
-	
+
 	if (elapsed_time_len == 2) {
 		const u16 var16 = htons((u16)elapsed_time);
 		memcpy(to, &var16, 2);
@@ -421,6 +458,93 @@ static void dccp_insert_option_timestamp_echo(struct sock *sk,
 	dp->dccps_timestamp_time.tv_usec = 0;
 }
 
+static int dccp_insert_feat_opt(struct sk_buff *skb, u8 type, u8 feat,
+			        u8 *val, u8 len)
+{
+	u8 *to;
+
+	if (DCCP_SKB_CB(skb)->dccpd_opt_len + len + 3 > DCCP_MAX_OPT_LEN) {
+		LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small"
+			       " to insert feature %d option!\n", feat);
+		return -1;
+	}
+
+	DCCP_SKB_CB(skb)->dccpd_opt_len += len + 3;
+
+	to    = skb_push(skb, len + 3);
+	*to++ = type;
+	*to++ = len + 3;
+	*to++ = feat;
+
+	if (len)
+		memcpy(to, val, len);
+	dccp_pr_debug("option %d feat %d len %d\n", type, feat, len);
+
+	return 0;
+}
+
+static void dccp_insert_feat(struct sock *sk, struct sk_buff *skb)
+{
+	struct dccp_sock *dp = dccp_sk(sk);
+	struct dccp_opt_pend *opt, *next;
+	int change = 0;
+
+	/* confirm any options [NN opts] */
+	list_for_each_entry_safe(opt, next, &dp->dccps_options.dccpo_conf,
+				 dccpop_node) {
+		dccp_insert_feat_opt(skb, opt->dccpop_type,
+				     opt->dccpop_feat, opt->dccpop_val,
+				     opt->dccpop_len);
+		/* fear empty confirms */
+		if (opt->dccpop_val)
+			kfree(opt->dccpop_val);
+		kfree(opt);
+	}
+	INIT_LIST_HEAD(&dp->dccps_options.dccpo_conf);
+
+	/* see which features we need to send */
+	list_for_each_entry(opt, &dp->dccps_options.dccpo_pending,
+			    dccpop_node) {
+		/* see if we need to send any confirm */
+		if (opt->dccpop_sc) {
+			dccp_insert_feat_opt(skb, opt->dccpop_type + 1,
+					     opt->dccpop_feat,
+					     opt->dccpop_sc->dccpoc_val,
+					     opt->dccpop_sc->dccpoc_len);
+
+			BUG_ON(!opt->dccpop_sc->dccpoc_val);
+			kfree(opt->dccpop_sc->dccpoc_val);
+			kfree(opt->dccpop_sc);
+			opt->dccpop_sc = NULL;
+		}
+
+		/* any option not confirmed, re-send it */
+		if (!opt->dccpop_conf) {
+			dccp_insert_feat_opt(skb, opt->dccpop_type,
+					     opt->dccpop_feat, opt->dccpop_val,
+					     opt->dccpop_len);
+			change++;
+		}
+	}
+
+	/* Retransmit timer.
+	 * If this is the master listening sock, we don't set a timer on it.  It
+	 * should be fine because if the dude doesn't receive our RESPONSE
+	 * [which will contain the CHANGE] he will send another REQUEST which
+	 * will "retrnasmit" the change.
+	 */
+	if (change && dp->dccps_role != DCCP_ROLE_LISTEN) {
+		dccp_pr_debug("reset feat negotiation timer %p\n", sk);
+
+		/* XXX don't reset the timer on re-transmissions.  I.e. reset it
+		 * only when sending new stuff i guess.  Currently the timer
+		 * never backs off because on re-transmission it just resets it!
+		 */
+		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
+					  inet_csk(sk)->icsk_rto, DCCP_RTO_MAX);
+	}
+}
+
 void dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 {
 	struct dccp_sock *dp = dccp_sk(sk);
@@ -447,6 +571,17 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb)
 		dp->dccps_hc_tx_insert_options = 0;
 	}
 
+	/* Feature negotiation */
+	switch(DCCP_SKB_CB(skb)->dccpd_type) {
+		/* Data packets can't do feat negotiation */
+	case DCCP_PKT_DATA:
+	case DCCP_PKT_DATAACK:
+		break;
+	default:
+		dccp_insert_feat(sk, skb);
+		break;
+	}
+
 	/* XXX: insert other options when appropriate */
 
 	if (DCCP_SKB_CB(skb)->dccpd_opt_len != 0) {

net/dccp/output.c

@@ -64,6 +64,10 @@ static int dccp_transmit_skb(struct sock *sk, struct sk_buff *skb)
 		case DCCP_PKT_DATAACK:
 			break;
 
+		case DCCP_PKT_REQUEST:
+			set_ack = 0;
+			/* fall through */
+
 		case DCCP_PKT_SYNC:
 		case DCCP_PKT_SYNCACK:
 			ackno = dcb->dccpd_seq;

net/dccp/proto.c

@@ -37,6 +37,7 @@
 
 #include "ccid.h"
 #include "dccp.h"
+#include "feat.h"
 
 DEFINE_SNMP_STAT(struct dccp_mib, dccp_statistics) __read_mostly;
 
@@ -255,6 +256,39 @@ static int dccp_setsockopt_service(struct sock *sk, const u32 service,
 	return 0;
 }
 
+/* byte 1 is feature.  the rest is the preference list */
+static int dccp_setsockopt_change(struct sock *sk, int type,
+				  struct dccp_so_feat __user *optval)
+{
+	struct dccp_so_feat opt;
+	u8 *val;
+	int rc;
+
+	if (copy_from_user(&opt, optval, sizeof(opt)))
+		return -EFAULT;
+
+	val = kmalloc(opt.dccpsf_len, GFP_KERNEL);
+	if (!val)
+		return -ENOMEM;
+
+	if (copy_from_user(val, opt.dccpsf_val, opt.dccpsf_len)) {
+		rc = -EFAULT;
+		goto out_free_val;
+	}
+
+	rc = dccp_feat_change(sk, type, opt.dccpsf_feat, val, opt.dccpsf_len,
+			      GFP_KERNEL);
+	if (rc)
+		goto out_free_val;
+
+out:
+	return rc;
+
+out_free_val:
+	kfree(val);
+	goto out;
+}
+
 int dccp_setsockopt(struct sock *sk, int level, int optname,
 		    char __user *optval, int optlen)
 {
@@ -284,6 +318,25 @@ int dccp_setsockopt(struct sock *sk, int level, int optname,
 	case DCCP_SOCKOPT_PACKET_SIZE:
 		dp->dccps_packet_size = val;
 		break;
+
+	case DCCP_SOCKOPT_CHANGE_L:
+		if (optlen != sizeof(struct dccp_so_feat))
+			err = -EINVAL;
+		else
+			err = dccp_setsockopt_change(sk, DCCPO_CHANGE_L,
+					             (struct dccp_so_feat *)
+						     optval);
+		break;
+
+	case DCCP_SOCKOPT_CHANGE_R:
+		if (optlen != sizeof(struct dccp_so_feat))
+			err = -EINVAL;
+		else
+			err = dccp_setsockopt_change(sk, DCCPO_CHANGE_R,
+						     (struct dccp_so_feat *)
+						     optval);
+		break;
+
 	default:
 		err = -ENOPROTOOPT;
 		break;

net/dccp/timer.c

@@ -141,6 +141,17 @@ static void dccp_retransmit_timer(struct sock *sk)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 
+	/* retransmit timer is used for feature negotiation throughout
+	 * connection.  In this case, no packet is re-transmitted, but rather an
+	 * ack is generated and pending changes are splaced into its options.
+	 */
+	if (sk->sk_send_head == NULL) {
+		dccp_pr_debug("feat negotiation retransmit timeout %p\n", sk);
+		if (sk->sk_state == DCCP_OPEN)
+			dccp_send_ack(sk);
+		goto backoff;
+	}
+
 	/*
 	 * sk->sk_send_head has to have one skb with
 	 * DCCP_SKB_CB(skb)->dccpd_type set to one of the retransmittable DCCP
@@ -177,6 +188,7 @@ static void dccp_retransmit_timer(struct sock *sk)
 		goto out;
 	}
 
+backoff:
 	icsk->icsk_backoff++;
 	icsk->icsk_retransmits++;