[DCCP]: Finish the TIMEWAIT minisock support

Using most of the infrastructure TCP uses, with a dccp_death_row,
etc. As per my current interpretation of the draft what we have with
this changeset seems to be all we need (or very close to it 8)).

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

3 files changed, 63 insertions, 15 deletions

diff --git a/include/linux/dccp.h b/include/linux/dccp.h
index 431d58923ba..3dccdd5108b 100644
--- a/include/linux/dccp.h
+++ b/include/linux/dccp.h
@@ -194,6 +194,7 @@ enum {
 #include <linux/workqueue.h>
 
 #include <net/inet_connection_sock.h>
+#include <net/inet_timewait_sock.h>
 #include <net/sock.h>
 #include <net/tcp_states.h>
 #include <net/tcp.h>
@@ -354,6 +355,8 @@ static inline struct dccp_request_sock *dccp_rsk(const struct request_sock *req)
 	return (struct dccp_request_sock *)req;
 }
 
+extern struct inet_timewait_death_row dccp_death_row;
+
 /* Read about the ECN nonce to see why it is 253 */
 #define DCCP_MAX_ACK_VECTOR_LEN 253
 
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index 6bccf4dd1e7..f6da9328221 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -97,7 +97,7 @@ static int __dccp_v4_check_established(struct sock *sk, const __u16 lport,
 		NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
 	} else if (tw != NULL) {
 		/* Silly. Should hash-dance instead... */
-		dccp_tw_deschedule(tw);
+		inet_twsk_deschedule(tw, &dccp_death_row);
 		NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
 
 		inet_twsk_put(tw);
@@ -201,7 +201,7 @@ ok:
  		spin_unlock(&head->lock);
 
  		if (tw != NULL) {
- 			dccp_tw_deschedule(tw);
+ 			inet_twsk_deschedule(tw, &dccp_death_row);
  			inet_twsk_put(tw);
  		}
 
@@ -1131,8 +1131,9 @@ int dccp_v4_rcv(struct sk_buff *skb)
 	 */
 	       
 	if (sk->sk_state == DCCP_TIME_WAIT) {
-		dccp_pr_debug("sk->sk_state == DCCP_TIME_WAIT: discard_and_relse\n");
-                goto discard_and_relse;
+		dccp_pr_debug("sk->sk_state == DCCP_TIME_WAIT: "
+			      "do_time_wait\n");
+                goto do_time_wait;
 	}
 
 	if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
@@ -1179,6 +1180,10 @@ discard_it:
 discard_and_relse:
 	sock_put(sk);
 	goto discard_it;
+
+do_time_wait:
+	inet_twsk_put((struct inet_timewait_sock *)sk);
+	goto no_dccp_socket;
 }
 
 static int dccp_v4_init_sock(struct sock *sk)
@@ -1290,5 +1295,5 @@ struct proto dccp_v4_prot = {
 	.max_header		= MAX_DCCP_HEADER,
 	.obj_size		= sizeof(struct dccp_sock),
 	.rsk_prot		= &dccp_request_sock_ops,
-	.twsk_obj_size		= sizeof(struct inet_timewait_sock), /* FIXME! create dccp_timewait_sock */
+	.twsk_obj_size		= sizeof(struct inet_timewait_sock),
 };
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index e498e389fcc..a6a0b270fb6 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -22,18 +22,58 @@
 #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)
 {
-	/* FIXME: Implement */
-	dccp_pr_debug("Want to help? Start here\n");
-	dccp_set_state(sk, state);
-}
+	struct inet_timewait_sock *tw = NULL;
 
-/* This is for handling early-kills of TIME_WAIT sockets. */
-void dccp_tw_deschedule(struct inet_timewait_sock *tw)
-{
-	dccp_pr_debug("Want to help? Start here\n");
-	__inet_twsk_kill(tw, &dccp_hashinfo);
+	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,
@@ -55,7 +95,7 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
 
 		newdp->dccps_hc_rx_ackpkts = NULL;
 		newdp->dccps_role = DCCP_ROLE_SERVER;
-		newicsk->icsk_rto = TCP_TIMEOUT_INIT;
+		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,