[linux-2.6-block.git] / net / dccp / minisocks.c

/*
 *  net/dccp/minisocks.c
 *
 *  An implementation of the DCCP protocol
 *  Arnaldo Carvalho de Melo <acme@conectiva.com.br>
 *
 *	This program is free software; you can redistribute it and/or
 *	modify it under the terms of the GNU General Public License
 *	as published by the Free Software Foundation; either version
 *	2 of the License, or (at your option) any later version.
 */

#include <linux/dccp.h>
#include <linux/gfp.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <linux/timer.h>

#include <net/sock.h>
#include <net/xfrm.h>
#include <net/inet_timewait_sock.h>

#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,
	.hashinfo	= &dccp_hashinfo,
};

EXPORT_SYMBOL_GPL(dccp_death_row);

void dccp_time_wait(struct sock *sk, int state, int timeo)
{
	struct inet_timewait_sock *tw;

	tw = inet_twsk_alloc(sk, &dccp_death_row, state);

	if (tw != NULL) {
		const struct inet_connection_sock *icsk = inet_csk(sk);
		const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);
#if IS_ENABLED(CONFIG_IPV6)
		if (tw->tw_family == PF_INET6) {
			tw->tw_v6_daddr = sk->sk_v6_daddr;
			tw->tw_v6_rcv_saddr = sk->sk_v6_rcv_saddr;
			tw->tw_ipv6only = sk->sk_ipv6only;
		}
#endif
		/* Linkage updates. */
		__inet_twsk_hashdance(tw, sk, &dccp_hashinfo);

		/* Get the TIME_WAIT timeout firing. */
		if (timeo < rto)
			timeo = rto;

		tw->tw_timeout = DCCP_TIMEWAIT_LEN;
		if (state == DCCP_TIME_WAIT)
			timeo = DCCP_TIMEWAIT_LEN;

		inet_twsk_schedule(tw, timeo);
		inet_twsk_put(tw);
	} else {
		/* Sorry, if we're out of memory, just CLOSE this
		 * socket up.  We've got bigger problems than
		 * non-graceful socket closings.
		 */
		DCCP_WARN("time wait bucket table overflow\n");
	}

	dccp_done(sk);
}

struct sock *dccp_create_openreq_child(struct sock *sk,
				       const struct request_sock *req,
				       const struct sk_buff *skb)
{
	/*
	 * Step 3: Process LISTEN state
	 *
	 *   (* Generate a new socket and switch to that socket *)
	 *   Set S := new socket for this port pair
	 */
	struct sock *newsk = inet_csk_clone_lock(sk, req, GFP_ATOMIC);

	if (newsk != NULL) {
		struct dccp_request_sock *dreq = dccp_rsk(req);
		struct inet_connection_sock *newicsk = inet_csk(newsk);
		struct dccp_sock *newdp = dccp_sk(newsk);

		newdp->dccps_role	    = DCCP_ROLE_SERVER;
		newdp->dccps_hc_rx_ackvec   = NULL;
		newdp->dccps_service_list   = NULL;
		newdp->dccps_service	    = dreq->dreq_service;
		newdp->dccps_timestamp_echo = dreq->dreq_timestamp_echo;
		newdp->dccps_timestamp_time = dreq->dreq_timestamp_time;
		newicsk->icsk_rto	    = DCCP_TIMEOUT_INIT;

		INIT_LIST_HEAD(&newdp->dccps_featneg);
		/*
		 * Step 3: Process LISTEN state
		 *
		 *    Choose S.ISS (initial seqno) or set from Init Cookies
		 *    Initialize S.GAR := S.ISS
		 *    Set S.ISR, S.GSR from packet (or Init Cookies)
		 *
		 *    Setting AWL/AWH and SWL/SWH happens as part of the feature
		 *    activation below, as these windows all depend on the local
		 *    and remote Sequence Window feature values (7.5.2).
		 */
		newdp->dccps_iss = dreq->dreq_iss;
		newdp->dccps_gss = dreq->dreq_gss;
		newdp->dccps_gar = newdp->dccps_iss;
		newdp->dccps_isr = dreq->dreq_isr;
		newdp->dccps_gsr = dreq->dreq_gsr;

		/*
		 * Activate features: initialise CCIDs, sequence windows etc.
		 */
		if (dccp_feat_activate_values(newsk, &dreq->dreq_featneg)) {
			/* It is still raw copy of parent, so invalidate
			 * destructor and make plain sk_free() */
			newsk->sk_destruct = NULL;
			sk_free(newsk);
			return NULL;
		}
		dccp_init_xmit_timers(newsk);

		DCCP_INC_STATS_BH(DCCP_MIB_PASSIVEOPENS);
	}
	return newsk;
}

EXPORT_SYMBOL_GPL(dccp_create_openreq_child);

/*
 * Process an incoming packet for RESPOND sockets represented
 * as an request_sock.
 */
struct sock *dccp_check_req(struct sock *sk, struct sk_buff *skb,
			    struct request_sock *req)
{
	struct sock *child = NULL;
	struct dccp_request_sock *dreq = dccp_rsk(req);

	/* Check for retransmitted REQUEST */
	if (dccp_hdr(skb)->dccph_type == DCCP_PKT_REQUEST) {

		if (after48(DCCP_SKB_CB(skb)->dccpd_seq, dreq->dreq_gsr)) {
			dccp_pr_debug("Retransmitted REQUEST\n");
			dreq->dreq_gsr = DCCP_SKB_CB(skb)->dccpd_seq;
			/*
			 * Send another RESPONSE packet
			 * To protect against Request floods, increment retrans
			 * counter (backoff, monitored by dccp_response_timer).
			 */
			inet_rtx_syn_ack(sk, req);
		}
		/* Network Duplicate, discard packet */
		return NULL;
	}

	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;

	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_ACK &&
	    dccp_hdr(skb)->dccph_type != DCCP_PKT_DATAACK)
		goto drop;

	/* Invalid ACK */
	if (!between48(DCCP_SKB_CB(skb)->dccpd_ack_seq,
				dreq->dreq_iss, dreq->dreq_gss)) {
		dccp_pr_debug("Invalid ACK number: ack_seq=%llu, "
			      "dreq_iss=%llu, dreq_gss=%llu\n",
			      (unsigned long long)
			      DCCP_SKB_CB(skb)->dccpd_ack_seq,
			      (unsigned long long) dreq->dreq_iss,
			      (unsigned long long) dreq->dreq_gss);
		goto drop;
	}

	if (dccp_parse_options(sk, dreq, skb))
		 goto drop;

	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
	if (child == NULL)
		goto listen_overflow;

	inet_csk_reqsk_queue_drop(sk, req);
	inet_csk_reqsk_queue_add(sk, req, child);
out:
	return child;
listen_overflow:
	dccp_pr_debug("listen_overflow!\n");
	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_TOO_BUSY;
drop:
	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
		req->rsk_ops->send_reset(sk, skb);

	inet_csk_reqsk_queue_drop(sk, req);
	goto out;
}

EXPORT_SYMBOL_GPL(dccp_check_req);

/*
 *  Queue segment on the new socket if the new socket is active,
 *  otherwise we just shortcircuit this and continue with
 *  the new socket.
 */
int dccp_child_process(struct sock *parent, struct sock *child,
		       struct sk_buff *skb)
{
	int ret = 0;
	const int state = child->sk_state;

	if (!sock_owned_by_user(child)) {
		ret = dccp_rcv_state_process(child, skb, dccp_hdr(skb),
					     skb->len);

		/* Wakeup parent, send SIGIO */
		if (state == DCCP_RESPOND && child->sk_state != state)
			parent->sk_data_ready(parent);
	} else {
		/* Alas, it is possible again, because we do lookup
		 * in main socket hash table and lock on listening
		 * socket does not protect us more.
		 */
		__sk_add_backlog(child, skb);
	}

	bh_unlock_sock(child);
	sock_put(child);
	return ret;
}

EXPORT_SYMBOL_GPL(dccp_child_process);

void dccp_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
			 struct request_sock *rsk)
{
	DCCP_BUG("DCCP-ACK packets are never sent in LISTEN/RESPOND state");
}

EXPORT_SYMBOL_GPL(dccp_reqsk_send_ack);

int dccp_reqsk_init(struct request_sock *req,
		    struct dccp_sock const *dp, struct sk_buff const *skb)
{
	struct dccp_request_sock *dreq = dccp_rsk(req);

	inet_rsk(req)->ir_rmt_port = dccp_hdr(skb)->dccph_sport;
	inet_rsk(req)->ir_num	   = ntohs(dccp_hdr(skb)->dccph_dport);
	inet_rsk(req)->acked	   = 0;
	dreq->dreq_timestamp_echo  = 0;

	/* inherit feature negotiation options from listening socket */
	return dccp_feat_clone_list(&dp->dccps_featneg, &dreq->dreq_featneg);
}

EXPORT_SYMBOL_GPL(dccp_reqsk_init);
Commit	Line	Data
	1	/*
	2	* net/dccp/minisocks.c
	3	*
	4	* An implementation of the DCCP protocol
	5	* Arnaldo Carvalho de Melo <acme@conectiva.com.br>
	6	*
	7	* This program is free software; you can redistribute it and/or
	8	* modify it under the terms of the GNU General Public License
	9	* as published by the Free Software Foundation; either version
	10	* 2 of the License, or (at your option) any later version.
	11	*/
	12
	13	#include <linux/dccp.h>
	14	#include <linux/gfp.h>
	15	#include <linux/kernel.h>
	16	#include <linux/skbuff.h>
	17	#include <linux/timer.h>
	18
	19	#include <net/sock.h>
	20	#include <net/xfrm.h>
	21	#include <net/inet_timewait_sock.h>
	22
	23	#include "ackvec.h"
	24	#include "ccid.h"
	25	#include "dccp.h"
	26	#include "feat.h"
	27
	28	struct inet_timewait_death_row dccp_death_row = {
	29	.sysctl_max_tw_buckets = NR_FILE * 2,
	30	.hashinfo = &dccp_hashinfo,
	31	};
	32
	33	EXPORT_SYMBOL_GPL(dccp_death_row);
	34
	35	void dccp_time_wait(struct sock *sk, int state, int timeo)
	36	{
	37	struct inet_timewait_sock *tw;
	38
	39	tw = inet_twsk_alloc(sk, &dccp_death_row, state);
	40
	41	if (tw != NULL) {
	42	const struct inet_connection_sock *icsk = inet_csk(sk);
	43	const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);
	44	#if IS_ENABLED(CONFIG_IPV6)
	45	if (tw->tw_family == PF_INET6) {
	46	tw->tw_v6_daddr = sk->sk_v6_daddr;
	47	tw->tw_v6_rcv_saddr = sk->sk_v6_rcv_saddr;
	48	tw->tw_ipv6only = sk->sk_ipv6only;
	49	}
	50	#endif
	51	/* Linkage updates. */
	52	__inet_twsk_hashdance(tw, sk, &dccp_hashinfo);
	53
	54	/* Get the TIME_WAIT timeout firing. */
	55	if (timeo < rto)
	56	timeo = rto;
	57
	58	tw->tw_timeout = DCCP_TIMEWAIT_LEN;
	59	if (state == DCCP_TIME_WAIT)
	60	timeo = DCCP_TIMEWAIT_LEN;
	61
	62	inet_twsk_schedule(tw, timeo);
	63	inet_twsk_put(tw);
	64	} else {
	65	/* Sorry, if we're out of memory, just CLOSE this
	66	* socket up. We've got bigger problems than
	67	* non-graceful socket closings.
	68	*/
	69	DCCP_WARN("time wait bucket table overflow\n");
	70	}
	71
	72	dccp_done(sk);
	73	}
	74
	75	struct sock dccp_create_openreq_child(struct sock sk,
	76	const struct request_sock *req,
	77	const struct sk_buff *skb)
	78	{
	79	/*
	80	* Step 3: Process LISTEN state
	81	*
	82	* (* Generate a new socket and switch to that socket *)
	83	* Set S := new socket for this port pair
	84	*/
	85	struct sock *newsk = inet_csk_clone_lock(sk, req, GFP_ATOMIC);
	86
	87	if (newsk != NULL) {
	88	struct dccp_request_sock *dreq = dccp_rsk(req);
	89	struct inet_connection_sock *newicsk = inet_csk(newsk);
	90	struct dccp_sock *newdp = dccp_sk(newsk);
	91
	92	newdp->dccps_role = DCCP_ROLE_SERVER;
	93	newdp->dccps_hc_rx_ackvec = NULL;
	94	newdp->dccps_service_list = NULL;
	95	newdp->dccps_service = dreq->dreq_service;
	96	newdp->dccps_timestamp_echo = dreq->dreq_timestamp_echo;
	97	newdp->dccps_timestamp_time = dreq->dreq_timestamp_time;
	98	newicsk->icsk_rto = DCCP_TIMEOUT_INIT;
	99
	100	INIT_LIST_HEAD(&newdp->dccps_featneg);
	101	/*
	102	* Step 3: Process LISTEN state
	103	*
	104	* Choose S.ISS (initial seqno) or set from Init Cookies
	105	* Initialize S.GAR := S.ISS
	106	* Set S.ISR, S.GSR from packet (or Init Cookies)
	107	*
	108	* Setting AWL/AWH and SWL/SWH happens as part of the feature
	109	* activation below, as these windows all depend on the local
	110	* and remote Sequence Window feature values (7.5.2).
	111	*/
	112	newdp->dccps_iss = dreq->dreq_iss;
	113	newdp->dccps_gss = dreq->dreq_gss;
	114	newdp->dccps_gar = newdp->dccps_iss;
	115	newdp->dccps_isr = dreq->dreq_isr;
	116	newdp->dccps_gsr = dreq->dreq_gsr;
	117
	118	/*
	119	* Activate features: initialise CCIDs, sequence windows etc.
	120	*/
	121	if (dccp_feat_activate_values(newsk, &dreq->dreq_featneg)) {
	122	/* It is still raw copy of parent, so invalidate
	123	* destructor and make plain sk_free() */
	124	newsk->sk_destruct = NULL;
	125	sk_free(newsk);
	126	return NULL;
	127	}
	128	dccp_init_xmit_timers(newsk);
	129
	130	DCCP_INC_STATS_BH(DCCP_MIB_PASSIVEOPENS);
	131	}
	132	return newsk;
	133	}
	134
	135	EXPORT_SYMBOL_GPL(dccp_create_openreq_child);
	136
	137	/*
	138	* Process an incoming packet for RESPOND sockets represented
	139	* as an request_sock.
	140	*/
	141	struct sock dccp_check_req(struct sock sk, struct sk_buff *skb,
	142	struct request_sock *req)
	143	{
	144	struct sock *child = NULL;
	145	struct dccp_request_sock *dreq = dccp_rsk(req);
	146
	147	/* Check for retransmitted REQUEST */
	148	if (dccp_hdr(skb)->dccph_type == DCCP_PKT_REQUEST) {
	149
	150	if (after48(DCCP_SKB_CB(skb)->dccpd_seq, dreq->dreq_gsr)) {
	151	dccp_pr_debug("Retransmitted REQUEST\n");
	152	dreq->dreq_gsr = DCCP_SKB_CB(skb)->dccpd_seq;
	153	/*
	154	* Send another RESPONSE packet
	155	* To protect against Request floods, increment retrans
	156	* counter (backoff, monitored by dccp_response_timer).
	157	*/
	158	inet_rtx_syn_ack(sk, req);
	159	}
	160	/* Network Duplicate, discard packet */
	161	return NULL;
	162	}
	163
	164	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;
	165
	166	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_ACK &&
	167	dccp_hdr(skb)->dccph_type != DCCP_PKT_DATAACK)
	168	goto drop;
	169
	170	/* Invalid ACK */
	171	if (!between48(DCCP_SKB_CB(skb)->dccpd_ack_seq,
	172	dreq->dreq_iss, dreq->dreq_gss)) {
	173	dccp_pr_debug("Invalid ACK number: ack_seq=%llu, "
	174	"dreq_iss=%llu, dreq_gss=%llu\n",
	175	(unsigned long long)
	176	DCCP_SKB_CB(skb)->dccpd_ack_seq,
	177	(unsigned long long) dreq->dreq_iss,
	178	(unsigned long long) dreq->dreq_gss);
	179	goto drop;
	180	}
	181
	182	if (dccp_parse_options(sk, dreq, skb))
	183	goto drop;
	184
	185	child = inet_csk(sk)->icsk_af_ops->syn_recv_sock(sk, skb, req, NULL);
	186	if (child == NULL)
	187	goto listen_overflow;
	188
	189	inet_csk_reqsk_queue_drop(sk, req);
	190	inet_csk_reqsk_queue_add(sk, req, child);
	191	out:
	192	return child;
	193	listen_overflow:
	194	dccp_pr_debug("listen_overflow!\n");
	195	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_TOO_BUSY;
	196	drop:
	197	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
	198	req->rsk_ops->send_reset(sk, skb);
	199
	200	inet_csk_reqsk_queue_drop(sk, req);
	201	goto out;
	202	}
	203
	204	EXPORT_SYMBOL_GPL(dccp_check_req);
	205
	206	/*
	207	* Queue segment on the new socket if the new socket is active,
	208	* otherwise we just shortcircuit this and continue with
	209	* the new socket.
	210	*/
	211	int dccp_child_process(struct sock parent, struct sock child,
	212	struct sk_buff *skb)
	213	{
	214	int ret = 0;
	215	const int state = child->sk_state;
	216
	217	if (!sock_owned_by_user(child)) {
	218	ret = dccp_rcv_state_process(child, skb, dccp_hdr(skb),
	219	skb->len);
	220
	221	/* Wakeup parent, send SIGIO */
	222	if (state == DCCP_RESPOND && child->sk_state != state)
	223	parent->sk_data_ready(parent);
	224	} else {
	225	/* Alas, it is possible again, because we do lookup
	226	* in main socket hash table and lock on listening
	227	* socket does not protect us more.
	228	*/
	229	__sk_add_backlog(child, skb);
	230	}
	231
	232	bh_unlock_sock(child);
	233	sock_put(child);
	234	return ret;
	235	}
	236
	237	EXPORT_SYMBOL_GPL(dccp_child_process);
	238
	239	void dccp_reqsk_send_ack(struct sock sk, struct sk_buff skb,
	240	struct request_sock *rsk)
	241	{
	242	DCCP_BUG("DCCP-ACK packets are never sent in LISTEN/RESPOND state");
	243	}
	244
	245	EXPORT_SYMBOL_GPL(dccp_reqsk_send_ack);
	246
	247	int dccp_reqsk_init(struct request_sock *req,
	248	struct dccp_sock const dp, struct sk_buff const skb)
	249	{
	250	struct dccp_request_sock *dreq = dccp_rsk(req);
	251
	252	inet_rsk(req)->ir_rmt_port = dccp_hdr(skb)->dccph_sport;
	253	inet_rsk(req)->ir_num = ntohs(dccp_hdr(skb)->dccph_dport);
	254	inet_rsk(req)->acked = 0;
	255	dreq->dreq_timestamp_echo = 0;
	256
	257	/* inherit feature negotiation options from listening socket */
	258	return dccp_feat_clone_list(&dp->dccps_featneg, &dreq->dreq_featneg);
	259	}
	260
	261	EXPORT_SYMBOL_GPL(dccp_reqsk_init);