[deliverable/linux.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/config.h>
#include <linux/dccp.h>
#include <linux/skbuff.h>
#include <linux/timer.h>

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

#include "ccid.h"
#include "dccp.h"

struct inet_timewait_death_row dccp_death_row = {
	.sysctl_max_tw_buckets = NR_FILE * 2,
	.period		= DCCP_TIMEWAIT_LEN / INET_TWDR_TWKILL_SLOTS,
	.death_lock	= SPIN_LOCK_UNLOCKED,
	.hashinfo	= &dccp_hashinfo,
	.tw_timer	= TIMER_INITIALIZER(inet_twdr_hangman, 0,
					    (unsigned long)&dccp_death_row),
	.twkill_work	= __WORK_INITIALIZER(dccp_death_row.twkill_work,
					     inet_twdr_twkill_work,
					     &dccp_death_row),
/* Short-time timewait calendar */

	.twcal_hand	= -1,
	.twcal_timer	= TIMER_INITIALIZER(inet_twdr_twcal_tick, 0,
					    (unsigned long)&dccp_death_row),
};

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

	if (dccp_death_row.tw_count < dccp_death_row.sysctl_max_tw_buckets)
		tw = inet_twsk_alloc(sk, state);

	if (tw != NULL) {
		const struct inet_connection_sock *icsk = inet_csk(sk);
		const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);

		/* 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, &dccp_death_row, timeo,
				   DCCP_TIMEWAIT_LEN);
		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.
		 */
		if (net_ratelimit())
			printk(KERN_INFO "DCCP: 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(sk, req, GFP_ATOMIC);

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

		newdp->dccps_hc_rx_ackpkts = NULL;
		newdp->dccps_role = DCCP_ROLE_SERVER;
		newicsk->icsk_rto = DCCP_TIMEOUT_INIT;

		if (newdp->dccps_options.dccpo_send_ack_vector) {
			newdp->dccps_hc_rx_ackpkts = dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN,
									GFP_ATOMIC);
			/*
			 * XXX: We're using the same CCIDs set on the parent, i.e. sk_clone
			 * copied the master sock and left the CCID pointers for this child,
			 * that is why we do the __ccid_get calls.
			 */
			if (unlikely(newdp->dccps_hc_rx_ackpkts == NULL))
				goto out_free;
		}

		if (unlikely(ccid_hc_rx_init(newdp->dccps_hc_rx_ccid, newsk) != 0 ||
			     ccid_hc_tx_init(newdp->dccps_hc_tx_ccid, newsk) != 0)) {
			dccp_ackpkts_free(newdp->dccps_hc_rx_ackpkts);
			ccid_hc_rx_exit(newdp->dccps_hc_rx_ccid, newsk);
			ccid_hc_tx_exit(newdp->dccps_hc_tx_ccid, newsk);
out_free:
			/* It is still raw copy of parent, so invalidate
			 * destructor and make plain sk_free() */
			newsk->sk_destruct = NULL;
			sk_free(newsk);
			return NULL;
		}

		__ccid_get(newdp->dccps_hc_rx_ccid);
		__ccid_get(newdp->dccps_hc_tx_ccid);

		/*
		 * Step 3: Process LISTEN state
		 *
		 *	Choose S.ISS (initial seqno) or set from Init Cookie
		 *	Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookie
		 */

		/* See dccp_v4_conn_request */
		newdp->dccps_options.dccpo_sequence_window = req->rcv_wnd;

		newdp->dccps_gar = newdp->dccps_isr = dreq->dreq_isr;
		dccp_update_gsr(newsk, dreq->dreq_isr);

		newdp->dccps_iss = dreq->dreq_iss;
		dccp_update_gss(newsk, dreq->dreq_iss);

		dccp_init_xmit_timers(newsk);

		DCCP_INC_STATS_BH(DCCP_MIB_PASSIVEOPENS);
	}
	return newsk;
}

/* 
 * 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 request_sock **prev)
{
	struct sock *child = NULL;

	/* Check for retransmitted REQUEST */
	if (dccp_hdr(skb)->dccph_type == DCCP_PKT_REQUEST) {
		if (after48(DCCP_SKB_CB(skb)->dccpd_seq, dccp_rsk(req)->dreq_isr)) {
			struct dccp_request_sock *dreq = dccp_rsk(req);

			dccp_pr_debug("Retransmitted REQUEST\n");
			/* Send another RESPONSE packet */
			dccp_set_seqno(&dreq->dreq_iss, dreq->dreq_iss + 1);
			dccp_set_seqno(&dreq->dreq_isr, DCCP_SKB_CB(skb)->dccpd_seq);
			req->rsk_ops->rtx_syn_ack(sk, req, NULL);
		}
		/* 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 (DCCP_SKB_CB(skb)->dccpd_ack_seq != dccp_rsk(req)->dreq_iss) {
		dccp_pr_debug("Invalid ACK number: ack_seq=%llu, dreq_iss=%llu\n",
			      (unsigned long long)
			      DCCP_SKB_CB(skb)->dccpd_ack_seq,
			      (unsigned long long)
			      dccp_rsk(req)->dreq_iss);
		goto drop;
	}

	child = dccp_v4_request_recv_sock(sk, skb, req, NULL);
	if (child == NULL)
		goto listen_overflow;

	/* FIXME: deal with options */

	inet_csk_reqsk_queue_unlink(sk, req, prev);
	inet_csk_reqsk_queue_removed(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(skb);

	inet_csk_reqsk_queue_drop(sk, req, prev);
	goto out;
}

/*
 *  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, 0);
	} 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;
}
Commit	Line	Data
7c657876 ACM	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/config.h>
	14	#include <linux/dccp.h>
	15	#include <linux/skbuff.h>
	16	#include <linux/timer.h>
	17
	18	#include <net/sock.h>
	19	#include <net/xfrm.h>
	20	#include <net/inet_timewait_sock.h>
	21
	22	#include "ccid.h"
	23	#include "dccp.h"
	24
64cf1e5d ACM	25	struct inet_timewait_death_row dccp_death_row = {
	26	.sysctl_max_tw_buckets = NR_FILE * 2,
	27	.period = DCCP_TIMEWAIT_LEN / INET_TWDR_TWKILL_SLOTS,
	28	.death_lock = SPIN_LOCK_UNLOCKED,
	29	.hashinfo = &dccp_hashinfo,
	30	.tw_timer = TIMER_INITIALIZER(inet_twdr_hangman, 0,
	31	(unsigned long)&dccp_death_row),
	32	.twkill_work = __WORK_INITIALIZER(dccp_death_row.twkill_work,
	33	inet_twdr_twkill_work,
	34	&dccp_death_row),
	35	/* Short-time timewait calendar */
	36
	37	.twcal_hand = -1,
	38	.twcal_timer = TIMER_INITIALIZER(inet_twdr_twcal_tick, 0,
	39	(unsigned long)&dccp_death_row),
	40	};
	41
7c657876 ACM	42	void dccp_time_wait(struct sock *sk, int state, int timeo)
7c657876 ACM	43	{
64cf1e5d	44	struct inet_timewait_sock *tw = NULL;
7c657876	45
64cf1e5d ACM	46	if (dccp_death_row.tw_count < dccp_death_row.sysctl_max_tw_buckets)
	47	tw = inet_twsk_alloc(sk, state);
	48
	49	if (tw != NULL) {
	50	const struct inet_connection_sock *icsk = inet_csk(sk);
	51	const int rto = (icsk->icsk_rto << 2) - (icsk->icsk_rto >> 1);
	52
	53	/* Linkage updates. */
	54	__inet_twsk_hashdance(tw, sk, &dccp_hashinfo);
	55
	56	/* Get the TIME_WAIT timeout firing. */
	57	if (timeo < rto)
	58	timeo = rto;
	59
	60	tw->tw_timeout = DCCP_TIMEWAIT_LEN;
	61	if (state == DCCP_TIME_WAIT)
	62	timeo = DCCP_TIMEWAIT_LEN;
	63
	64	inet_twsk_schedule(tw, &dccp_death_row, timeo,
	65	DCCP_TIMEWAIT_LEN);
	66	inet_twsk_put(tw);
	67	} else {
	68	/* Sorry, if we're out of memory, just CLOSE this
	69	* socket up. We've got bigger problems than
	70	* non-graceful socket closings.
	71	*/
	72	if (net_ratelimit())
	73	printk(KERN_INFO "DCCP: time wait bucket table overflow\n");
	74	}
	75
	76	dccp_done(sk);
7c657876 ACM	77	}
	78
	79	struct sock dccp_create_openreq_child(struct sock sk,
	80	const struct request_sock *req,
	81	const struct sk_buff *skb)
	82	{
	83	/*
	84	* Step 3: Process LISTEN state
	85	*
	86	* // Generate a new socket and switch to that socket
	87	* Set S := new socket for this port pair
	88	*/
	89	struct sock *newsk = inet_csk_clone(sk, req, GFP_ATOMIC);
	90
	91	if (newsk != NULL) {
	92	const struct dccp_request_sock *dreq = dccp_rsk(req);
	93	struct inet_connection_sock *newicsk = inet_csk(sk);
	94	struct dccp_sock *newdp = dccp_sk(newsk);
	95
	96	newdp->dccps_hc_rx_ackpkts = NULL;
	97	newdp->dccps_role = DCCP_ROLE_SERVER;
64cf1e5d	98	newicsk->icsk_rto = DCCP_TIMEOUT_INIT;
7c657876 ACM	99
	100	if (newdp->dccps_options.dccpo_send_ack_vector) {
	101	newdp->dccps_hc_rx_ackpkts = dccp_ackpkts_alloc(DCCP_MAX_ACK_VECTOR_LEN,
	102	GFP_ATOMIC);
	103	/*
	104	* XXX: We're using the same CCIDs set on the parent, i.e. sk_clone
	105	* copied the master sock and left the CCID pointers for this child,
	106	* that is why we do the __ccid_get calls.
	107	*/
	108	if (unlikely(newdp->dccps_hc_rx_ackpkts == NULL))
	109	goto out_free;
	110	}
	111
	112	if (unlikely(ccid_hc_rx_init(newdp->dccps_hc_rx_ccid, newsk) != 0 \|\|
	113	ccid_hc_tx_init(newdp->dccps_hc_tx_ccid, newsk) != 0)) {
	114	dccp_ackpkts_free(newdp->dccps_hc_rx_ackpkts);
	115	ccid_hc_rx_exit(newdp->dccps_hc_rx_ccid, newsk);
	116	ccid_hc_tx_exit(newdp->dccps_hc_tx_ccid, newsk);
	117	out_free:
	118	/* It is still raw copy of parent, so invalidate
	119	* destructor and make plain sk_free() */
	120	newsk->sk_destruct = NULL;
	121	sk_free(newsk);
	122	return NULL;
	123	}
	124
	125	__ccid_get(newdp->dccps_hc_rx_ccid);
	126	__ccid_get(newdp->dccps_hc_tx_ccid);
	127
	128	/*
	129	* Step 3: Process LISTEN state
	130	*
	131	* Choose S.ISS (initial seqno) or set from Init Cookie
	132	* Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookie
	133	*/
	134
	135	/* See dccp_v4_conn_request */
	136	newdp->dccps_options.dccpo_sequence_window = req->rcv_wnd;
	137
	138	newdp->dccps_gar = newdp->dccps_isr = dreq->dreq_isr;
	139	dccp_update_gsr(newsk, dreq->dreq_isr);
	140
	141	newdp->dccps_iss = dreq->dreq_iss;
	142	dccp_update_gss(newsk, dreq->dreq_iss);
	143
	144	dccp_init_xmit_timers(newsk);
	145
	146	DCCP_INC_STATS_BH(DCCP_MIB_PASSIVEOPENS);
	147	}
	148	return newsk;
	149	}
	150
	151	/*
	152	* Process an incoming packet for RESPOND sockets represented
	153	* as an request_sock.
	154	*/
	155	struct sock dccp_check_req(struct sock sk, struct sk_buff *skb,
	156	struct request_sock *req,
	157	struct request_sock **prev)
	158	{
	159	struct sock *child = NULL;
	160
	161	/* Check for retransmitted REQUEST */
	162	if (dccp_hdr(skb)->dccph_type == DCCP_PKT_REQUEST) {
163	if (after48(DCCP_SKB_CB(skb)->dccpd_seq, dccp_rsk(req)->dreq_isr)) {
164	struct dccp_request_sock *dreq = dccp_rsk(req);
165
166	dccp_pr_debug("Retransmitted REQUEST\n");
167	/* Send another RESPONSE packet */
168	dccp_set_seqno(&dreq->dreq_iss, dreq->dreq_iss + 1);
169	dccp_set_seqno(&dreq->dreq_isr, DCCP_SKB_CB(skb)->dccpd_seq);
170	req->rsk_ops->rtx_syn_ack(sk, req, NULL);
171	}
172	/* Network Duplicate, discard packet */
173	return NULL;
174	}
175
176	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;
177
178	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_ACK &&
179	dccp_hdr(skb)->dccph_type != DCCP_PKT_DATAACK)
180	goto drop;
181
182	/* Invalid ACK */
183	if (DCCP_SKB_CB(skb)->dccpd_ack_seq != dccp_rsk(req)->dreq_iss) {
184	dccp_pr_debug("Invalid ACK number: ack_seq=%llu, dreq_iss=%llu\n",
f6ccf554 DM	185	(unsigned long long)
	186	DCCP_SKB_CB(skb)->dccpd_ack_seq,
	187	(unsigned long long)
	188	dccp_rsk(req)->dreq_iss);
7c657876 ACM	189	goto drop;
	190	}
	191
	192	child = dccp_v4_request_recv_sock(sk, skb, req, NULL);
	193	if (child == NULL)
	194	goto listen_overflow;
	195
	196	/* FIXME: deal with options */
	197
	198	inet_csk_reqsk_queue_unlink(sk, req, prev);
	199	inet_csk_reqsk_queue_removed(sk, req);
	200	inet_csk_reqsk_queue_add(sk, req, child);
	201	out:
	202	return child;
	203	listen_overflow:
	204	dccp_pr_debug("listen_overflow!\n");
	205	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_TOO_BUSY;
	206	drop:
	207	if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
	208	req->rsk_ops->send_reset(skb);
	209
	210	inet_csk_reqsk_queue_drop(sk, req, prev);
	211	goto out;
	212	}
	213
	214	/*
	215	* Queue segment on the new socket if the new socket is active,
	216	* otherwise we just shortcircuit this and continue with
	217	* the new socket.
	218	*/
	219	int dccp_child_process(struct sock parent, struct sock child,
	220	struct sk_buff *skb)
	221	{
	222	int ret = 0;
	223	const int state = child->sk_state;
	224
	225	if (!sock_owned_by_user(child)) {
	226	ret = dccp_rcv_state_process(child, skb, dccp_hdr(skb), skb->len);
	227
	228	/* Wakeup parent, send SIGIO */
	229	if (state == DCCP_RESPOND && child->sk_state != state)
	230	parent->sk_data_ready(parent, 0);
	231	} else {
	232	/* Alas, it is possible again, because we do lookup
	233	* in main socket hash table and lock on listening
	234	* socket does not protect us more.
	235	*/
	236	sk_add_backlog(child, skb);
	237	}
	238
	239	bh_unlock_sock(child);
	240	sock_put(child);
	241	return ret;
	242	}