[linux-2.6-block.git] / net / rxrpc / peer_event.c

// SPDX-License-Identifier: GPL-2.0-or-later
/* Peer event handling, typically ICMP messages.
 *
 * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
 * Written by David Howells (dhowells@redhat.com)
 */

#include <linux/module.h>
#include <linux/net.h>
#include <linux/skbuff.h>
#include <linux/errqueue.h>
#include <linux/udp.h>
#include <linux/in.h>
#include <linux/in6.h>
#include <linux/icmp.h>
#include <net/sock.h>
#include <net/af_rxrpc.h>
#include <net/ip.h>
#include "ar-internal.h"

static void rxrpc_store_error(struct rxrpc_peer *, struct sock_exterr_skb *);
static void rxrpc_distribute_error(struct rxrpc_peer *, int,
				   enum rxrpc_call_completion);

/*
 * Find the peer associated with an ICMP packet.
 */
static struct rxrpc_peer *rxrpc_lookup_peer_icmp_rcu(struct rxrpc_local *local,
						     const struct sk_buff *skb,
						     struct sockaddr_rxrpc *srx)
{
	struct sock_exterr_skb *serr = SKB_EXT_ERR(skb);

	_enter("");

	memset(srx, 0, sizeof(*srx));
	srx->transport_type = local->srx.transport_type;
	srx->transport_len = local->srx.transport_len;
	srx->transport.family = local->srx.transport.family;

	/* Can we see an ICMP4 packet on an ICMP6 listening socket?  and vice
	 * versa?
	 */
	switch (srx->transport.family) {
	case AF_INET:
		srx->transport_len = sizeof(srx->transport.sin);
		srx->transport.family = AF_INET;
		srx->transport.sin.sin_port = serr->port;
		switch (serr->ee.ee_origin) {
		case SO_EE_ORIGIN_ICMP:
			_net("Rx ICMP");
			memcpy(&srx->transport.sin.sin_addr,
			       skb_network_header(skb) + serr->addr_offset,
			       sizeof(struct in_addr));
			break;
		case SO_EE_ORIGIN_ICMP6:
			_net("Rx ICMP6 on v4 sock");
			memcpy(&srx->transport.sin.sin_addr,
			       skb_network_header(skb) + serr->addr_offset + 12,
			       sizeof(struct in_addr));
			break;
		default:
			memcpy(&srx->transport.sin.sin_addr, &ip_hdr(skb)->saddr,
			       sizeof(struct in_addr));
			break;
		}
		break;

#ifdef CONFIG_AF_RXRPC_IPV6
	case AF_INET6:
		switch (serr->ee.ee_origin) {
		case SO_EE_ORIGIN_ICMP6:
			_net("Rx ICMP6");
			srx->transport.sin6.sin6_port = serr->port;
			memcpy(&srx->transport.sin6.sin6_addr,
			       skb_network_header(skb) + serr->addr_offset,
			       sizeof(struct in6_addr));
			break;
		case SO_EE_ORIGIN_ICMP:
			_net("Rx ICMP on v6 sock");
			srx->transport_len = sizeof(srx->transport.sin);
			srx->transport.family = AF_INET;
			srx->transport.sin.sin_port = serr->port;
			memcpy(&srx->transport.sin.sin_addr,
			       skb_network_header(skb) + serr->addr_offset,
			       sizeof(struct in_addr));
			break;
		default:
			memcpy(&srx->transport.sin6.sin6_addr,
			       &ipv6_hdr(skb)->saddr,
			       sizeof(struct in6_addr));
			break;
		}
		break;
#endif

	default:
		BUG();
	}

	return rxrpc_lookup_peer_rcu(local, srx);
}

/*
 * Handle an MTU/fragmentation problem.
 */
static void rxrpc_adjust_mtu(struct rxrpc_peer *peer, struct sock_exterr_skb *serr)
{
	u32 mtu = serr->ee.ee_info;

	_net("Rx ICMP Fragmentation Needed (%d)", mtu);

	/* wind down the local interface MTU */
	if (mtu > 0 && peer->if_mtu == 65535 && mtu < peer->if_mtu) {
		peer->if_mtu = mtu;
		_net("I/F MTU %u", mtu);
	}

	if (mtu == 0) {
		/* they didn't give us a size, estimate one */
		mtu = peer->if_mtu;
		if (mtu > 1500) {
			mtu >>= 1;
			if (mtu < 1500)
				mtu = 1500;
		} else {
			mtu -= 100;
			if (mtu < peer->hdrsize)
				mtu = peer->hdrsize + 4;
		}
	}

	if (mtu < peer->mtu) {
		spin_lock_bh(&peer->lock);
		peer->mtu = mtu;
		peer->maxdata = peer->mtu - peer->hdrsize;
		spin_unlock_bh(&peer->lock);
		_net("Net MTU %u (maxdata %u)",
		     peer->mtu, peer->maxdata);
	}
}

/*
 * Handle an error received on the local endpoint.
 */
void rxrpc_error_report(struct sock *sk)
{
	struct sock_exterr_skb *serr;
	struct sockaddr_rxrpc srx;
	struct rxrpc_local *local;
	struct rxrpc_peer *peer;
	struct sk_buff *skb;

	rcu_read_lock();
	local = rcu_dereference_sk_user_data(sk);
	if (unlikely(!local)) {
		rcu_read_unlock();
		return;
	}
	_enter("%p{%d}", sk, local->debug_id);

	/* Clear the outstanding error value on the socket so that it doesn't
	 * cause kernel_sendmsg() to return it later.
	 */
	sock_error(sk);

	skb = sock_dequeue_err_skb(sk);
	if (!skb) {
		rcu_read_unlock();
		_leave("UDP socket errqueue empty");
		return;
	}
	rxrpc_new_skb(skb, rxrpc_skb_received);
	serr = SKB_EXT_ERR(skb);
	if (!skb->len && serr->ee.ee_origin == SO_EE_ORIGIN_TIMESTAMPING) {
		_leave("UDP empty message");
		rcu_read_unlock();
		rxrpc_free_skb(skb, rxrpc_skb_freed);
		return;
	}

	peer = rxrpc_lookup_peer_icmp_rcu(local, skb, &srx);
	if (peer && !rxrpc_get_peer_maybe(peer))
		peer = NULL;
	if (!peer) {
		rcu_read_unlock();
		rxrpc_free_skb(skb, rxrpc_skb_freed);
		_leave(" [no peer]");
		return;
	}

	trace_rxrpc_rx_icmp(peer, &serr->ee, &srx);

	if ((serr->ee.ee_origin == SO_EE_ORIGIN_ICMP &&
	     serr->ee.ee_type == ICMP_DEST_UNREACH &&
	     serr->ee.ee_code == ICMP_FRAG_NEEDED)) {
		rxrpc_adjust_mtu(peer, serr);
		rcu_read_unlock();
		rxrpc_free_skb(skb, rxrpc_skb_freed);
		rxrpc_put_peer(peer);
		_leave(" [MTU update]");
		return;
	}

	rxrpc_store_error(peer, serr);
	rcu_read_unlock();
	rxrpc_free_skb(skb, rxrpc_skb_freed);
	rxrpc_put_peer(peer);

	_leave("");
}

/*
 * Map an error report to error codes on the peer record.
 */
static void rxrpc_store_error(struct rxrpc_peer *peer,
			      struct sock_exterr_skb *serr)
{
	enum rxrpc_call_completion compl = RXRPC_CALL_NETWORK_ERROR;
	struct sock_extended_err *ee;
	int err;

	_enter("");

	ee = &serr->ee;

	err = ee->ee_errno;

	switch (ee->ee_origin) {
	case SO_EE_ORIGIN_ICMP:
		switch (ee->ee_type) {
		case ICMP_DEST_UNREACH:
			switch (ee->ee_code) {
			case ICMP_NET_UNREACH:
				_net("Rx Received ICMP Network Unreachable");
				break;
			case ICMP_HOST_UNREACH:
				_net("Rx Received ICMP Host Unreachable");
				break;
			case ICMP_PORT_UNREACH:
				_net("Rx Received ICMP Port Unreachable");
				break;
			case ICMP_NET_UNKNOWN:
				_net("Rx Received ICMP Unknown Network");
				break;
			case ICMP_HOST_UNKNOWN:
				_net("Rx Received ICMP Unknown Host");
				break;
			default:
				_net("Rx Received ICMP DestUnreach code=%u",
				     ee->ee_code);
				break;
			}
			break;

		case ICMP_TIME_EXCEEDED:
			_net("Rx Received ICMP TTL Exceeded");
			break;

		default:
			_proto("Rx Received ICMP error { type=%u code=%u }",
			       ee->ee_type, ee->ee_code);
			break;
		}
		break;

	case SO_EE_ORIGIN_NONE:
	case SO_EE_ORIGIN_LOCAL:
		_proto("Rx Received local error { error=%d }", err);
		compl = RXRPC_CALL_LOCAL_ERROR;
		break;

	case SO_EE_ORIGIN_ICMP6:
	default:
		_proto("Rx Received error report { orig=%u }", ee->ee_origin);
		break;
	}

	rxrpc_distribute_error(peer, err, compl);
}

/*
 * Distribute an error that occurred on a peer.
 */
static void rxrpc_distribute_error(struct rxrpc_peer *peer, int error,
				   enum rxrpc_call_completion compl)
{
	struct rxrpc_call *call;

	hlist_for_each_entry_rcu(call, &peer->error_targets, error_link) {
		rxrpc_see_call(call);
		if (call->state < RXRPC_CALL_COMPLETE &&
		    rxrpc_set_call_completion(call, compl, 0, -error))
			rxrpc_notify_socket(call);
	}
}

/*
 * Add RTT information to cache.  This is called in softirq mode and has
 * exclusive access to the peer RTT data.
 */
void rxrpc_peer_add_rtt(struct rxrpc_call *call, enum rxrpc_rtt_rx_trace why,
			rxrpc_serial_t send_serial, rxrpc_serial_t resp_serial,
			ktime_t send_time, ktime_t resp_time)
{
	struct rxrpc_peer *peer = call->peer;
	s64 rtt;
	u64 sum = peer->rtt_sum, avg;
	u8 cursor = peer->rtt_cursor, usage = peer->rtt_usage;

	rtt = ktime_to_ns(ktime_sub(resp_time, send_time));
	if (rtt < 0)
		return;

	spin_lock(&peer->rtt_input_lock);

	/* Replace the oldest datum in the RTT buffer */
	sum -= peer->rtt_cache[cursor];
	sum += rtt;
	peer->rtt_cache[cursor] = rtt;
	peer->rtt_cursor = (cursor + 1) & (RXRPC_RTT_CACHE_SIZE - 1);
	peer->rtt_sum = sum;
	if (usage < RXRPC_RTT_CACHE_SIZE) {
		usage++;
		peer->rtt_usage = usage;
	}

	spin_unlock(&peer->rtt_input_lock);

	/* Now recalculate the average */
	if (usage == RXRPC_RTT_CACHE_SIZE) {
		avg = sum / RXRPC_RTT_CACHE_SIZE;
	} else {
		avg = sum;
		do_div(avg, usage);
	}

	/* Don't need to update this under lock */
	peer->rtt = avg;
	trace_rxrpc_rtt_rx(call, why, send_serial, resp_serial, rtt,
			   usage, avg);
}

/*
 * Perform keep-alive pings.
 */
static void rxrpc_peer_keepalive_dispatch(struct rxrpc_net *rxnet,
					  struct list_head *collector,
					  time64_t base,
					  u8 cursor)
{
	struct rxrpc_peer *peer;
	const u8 mask = ARRAY_SIZE(rxnet->peer_keepalive) - 1;
	time64_t keepalive_at;
	int slot;

	spin_lock_bh(&rxnet->peer_hash_lock);

	while (!list_empty(collector)) {
		peer = list_entry(collector->next,
				  struct rxrpc_peer, keepalive_link);

		list_del_init(&peer->keepalive_link);
		if (!rxrpc_get_peer_maybe(peer))
			continue;

		spin_unlock_bh(&rxnet->peer_hash_lock);

		keepalive_at = peer->last_tx_at + RXRPC_KEEPALIVE_TIME;
		slot = keepalive_at - base;
		_debug("%02x peer %u t=%d {%pISp}",
		       cursor, peer->debug_id, slot, &peer->srx.transport);

		if (keepalive_at <= base ||
		    keepalive_at > base + RXRPC_KEEPALIVE_TIME) {
			rxrpc_send_keepalive(peer);
			slot = RXRPC_KEEPALIVE_TIME;
		}

		/* A transmission to this peer occurred since last we examined
		 * it so put it into the appropriate future bucket.
		 */
		slot += cursor;
		slot &= mask;
		spin_lock_bh(&rxnet->peer_hash_lock);
		list_add_tail(&peer->keepalive_link,
			      &rxnet->peer_keepalive[slot & mask]);
		rxrpc_put_peer_locked(peer);
	}

	spin_unlock_bh(&rxnet->peer_hash_lock);
}

/*
 * Perform keep-alive pings with VERSION packets to keep any NAT alive.
 */
void rxrpc_peer_keepalive_worker(struct work_struct *work)
{
	struct rxrpc_net *rxnet =
		container_of(work, struct rxrpc_net, peer_keepalive_work);
	const u8 mask = ARRAY_SIZE(rxnet->peer_keepalive) - 1;
	time64_t base, now, delay;
	u8 cursor, stop;
	LIST_HEAD(collector);

	now = ktime_get_seconds();
	base = rxnet->peer_keepalive_base;
	cursor = rxnet->peer_keepalive_cursor;
	_enter("%lld,%u", base - now, cursor);

	if (!rxnet->live)
		return;

	/* Remove to a temporary list all the peers that are currently lodged
	 * in expired buckets plus all new peers.
	 *
	 * Everything in the bucket at the cursor is processed this
	 * second; the bucket at cursor + 1 goes at now + 1s and so
	 * on...
	 */
	spin_lock_bh(&rxnet->peer_hash_lock);
	list_splice_init(&rxnet->peer_keepalive_new, &collector);

	stop = cursor + ARRAY_SIZE(rxnet->peer_keepalive);
	while (base <= now && (s8)(cursor - stop) < 0) {
		list_splice_tail_init(&rxnet->peer_keepalive[cursor & mask],
				      &collector);
		base++;
		cursor++;
	}

	base = now;
	spin_unlock_bh(&rxnet->peer_hash_lock);

	rxnet->peer_keepalive_base = base;
	rxnet->peer_keepalive_cursor = cursor;
	rxrpc_peer_keepalive_dispatch(rxnet, &collector, base, cursor);
	ASSERT(list_empty(&collector));

	/* Schedule the timer for the next occupied timeslot. */
	cursor = rxnet->peer_keepalive_cursor;
	stop = cursor + RXRPC_KEEPALIVE_TIME - 1;
	for (; (s8)(cursor - stop) < 0; cursor++) {
		if (!list_empty(&rxnet->peer_keepalive[cursor & mask]))
			break;
		base++;
	}

	now = ktime_get_seconds();
	delay = base - now;
	if (delay < 1)
		delay = 1;
	delay *= HZ;
	if (rxnet->live)
		timer_reduce(&rxnet->peer_keepalive_timer, jiffies + delay);

	_leave("");
}
Commit	Line	Data
2874c5fd	1	// SPDX-License-Identifier: GPL-2.0-or-later
f66d7490	2	/* Peer event handling, typically ICMP messages.
17926a79 DH	3	*
	4	* Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
	5	* Written by David Howells (dhowells@redhat.com)
17926a79 DH	6	*/
	7
	8	#include <linux/module.h>
	9	#include <linux/net.h>
	10	#include <linux/skbuff.h>
	11	#include <linux/errqueue.h>
	12	#include <linux/udp.h>
	13	#include <linux/in.h>
	14	#include <linux/in6.h>
	15	#include <linux/icmp.h>
	16	#include <net/sock.h>
	17	#include <net/af_rxrpc.h>
	18	#include <net/ip.h>
	19	#include "ar-internal.h"
	20
f66d7490	21	static void rxrpc_store_error(struct rxrpc_peer , struct sock_exterr_skb );
f3344303 DH	22	static void rxrpc_distribute_error(struct rxrpc_peer *, int,
f3344303 DH	23	enum rxrpc_call_completion);
f66d7490	24
be6e6707 DH	25	/*
	26	* Find the peer associated with an ICMP packet.
	27	*/
	28	static struct rxrpc_peer rxrpc_lookup_peer_icmp_rcu(struct rxrpc_local local,
494337c9 DH	29	const struct sk_buff *skb,
494337c9 DH	30	struct sockaddr_rxrpc *srx)
be6e6707 DH	31	{
be6e6707 DH	32	struct sock_exterr_skb *serr = SKB_EXT_ERR(skb);
be6e6707 DH	33
	34	_enter("");
	35
494337c9 DH	36	memset(srx, 0, sizeof(*srx));
	37	srx->transport_type = local->srx.transport_type;
	38	srx->transport_len = local->srx.transport_len;
	39	srx->transport.family = local->srx.transport.family;
be6e6707 DH	40
	41	/* Can we see an ICMP4 packet on an ICMP6 listening socket? and vice
	42	* versa?
	43	*/
494337c9	44	switch (srx->transport.family) {
be6e6707	45	case AF_INET:
46894a13 DH	46	srx->transport_len = sizeof(srx->transport.sin);
46894a13 DH	47	srx->transport.family = AF_INET;
494337c9	48	srx->transport.sin.sin_port = serr->port;
be6e6707 DH	49	switch (serr->ee.ee_origin) {
	50	case SO_EE_ORIGIN_ICMP:
	51	_net("Rx ICMP");
494337c9	52	memcpy(&srx->transport.sin.sin_addr,
be6e6707 DH	53	skb_network_header(skb) + serr->addr_offset,
	54	sizeof(struct in_addr));
	55	break;
	56	case SO_EE_ORIGIN_ICMP6:
	57	_net("Rx ICMP6 on v4 sock");
494337c9	58	memcpy(&srx->transport.sin.sin_addr,
be6e6707 DH	59	skb_network_header(skb) + serr->addr_offset + 12,
	60	sizeof(struct in_addr));
	61	break;
	62	default:
494337c9	63	memcpy(&srx->transport.sin.sin_addr, &ip_hdr(skb)->saddr,
be6e6707 DH	64	sizeof(struct in_addr));
	65	break;
	66	}
	67	break;
	68
d1912747	69	#ifdef CONFIG_AF_RXRPC_IPV6
75b54cb5	70	case AF_INET6:
75b54cb5 DH	71	switch (serr->ee.ee_origin) {
	72	case SO_EE_ORIGIN_ICMP6:
	73	_net("Rx ICMP6");
46894a13	74	srx->transport.sin6.sin6_port = serr->port;
494337c9	75	memcpy(&srx->transport.sin6.sin6_addr,
75b54cb5 DH	76	skb_network_header(skb) + serr->addr_offset,
	77	sizeof(struct in6_addr));
	78	break;
	79	case SO_EE_ORIGIN_ICMP:
	80	_net("Rx ICMP on v6 sock");
46894a13 DH	81	srx->transport_len = sizeof(srx->transport.sin);
	82	srx->transport.family = AF_INET;
	83	srx->transport.sin.sin_port = serr->port;
	84	memcpy(&srx->transport.sin.sin_addr,
75b54cb5 DH	85	skb_network_header(skb) + serr->addr_offset,
	86	sizeof(struct in_addr));
	87	break;
	88	default:
494337c9	89	memcpy(&srx->transport.sin6.sin6_addr,
75b54cb5 DH	90	&ipv6_hdr(skb)->saddr,
	91	sizeof(struct in6_addr));
	92	break;
	93	}
	94	break;
d1912747	95	#endif
75b54cb5	96
be6e6707 DH	97	default:
	98	BUG();
	99	}
	100
494337c9	101	return rxrpc_lookup_peer_rcu(local, srx);
be6e6707 DH	102	}
be6e6707 DH	103
1a70c05b DH	104	/*
	105	* Handle an MTU/fragmentation problem.
	106	*/
	107	static void rxrpc_adjust_mtu(struct rxrpc_peer peer, struct sock_exterr_skb serr)
	108	{
	109	u32 mtu = serr->ee.ee_info;
	110
	111	_net("Rx ICMP Fragmentation Needed (%d)", mtu);
	112
	113	/* wind down the local interface MTU */
	114	if (mtu > 0 && peer->if_mtu == 65535 && mtu < peer->if_mtu) {
	115	peer->if_mtu = mtu;
	116	_net("I/F MTU %u", mtu);
	117	}
	118
	119	if (mtu == 0) {
	120	/* they didn't give us a size, estimate one */
	121	mtu = peer->if_mtu;
	122	if (mtu > 1500) {
	123	mtu >>= 1;
	124	if (mtu < 1500)
	125	mtu = 1500;
	126	} else {
	127	mtu -= 100;
	128	if (mtu < peer->hdrsize)
	129	mtu = peer->hdrsize + 4;
	130	}
	131	}
	132
	133	if (mtu < peer->mtu) {
	134	spin_lock_bh(&peer->lock);
	135	peer->mtu = mtu;
	136	peer->maxdata = peer->mtu - peer->hdrsize;
	137	spin_unlock_bh(&peer->lock);
	138	_net("Net MTU %u (maxdata %u)",
	139	peer->mtu, peer->maxdata);
	140	}
	141	}
	142
17926a79	143	/*
f66d7490	144	* Handle an error received on the local endpoint.
17926a79	145	*/
abe89ef0	146	void rxrpc_error_report(struct sock *sk)
17926a79 DH	147	{
17926a79 DH	148	struct sock_exterr_skb *serr;
494337c9	149	struct sockaddr_rxrpc srx;
2ca4f6ca	150	struct rxrpc_local *local;
17926a79 DH	151	struct rxrpc_peer *peer;
17926a79 DH	152	struct sk_buff *skb;
17926a79	153
2ca4f6ca ED	154	rcu_read_lock();
	155	local = rcu_dereference_sk_user_data(sk);
	156	if (unlikely(!local)) {
	157	rcu_read_unlock();
f0308fb0	158	return;
2ca4f6ca	159	}
17926a79 DH	160	_enter("%p{%d}", sk, local->debug_id);
17926a79 DH	161
56d282d9 MD	162	/* Clear the outstanding error value on the socket so that it doesn't
	163	* cause kernel_sendmsg() to return it later.
	164	*/
	165	sock_error(sk);
	166
364a9e93	167	skb = sock_dequeue_err_skb(sk);
17926a79	168	if (!skb) {
2ca4f6ca	169	rcu_read_unlock();
17926a79 DH	170	_leave("UDP socket errqueue empty");
	171	return;
	172	}
987db9f7	173	rxrpc_new_skb(skb, rxrpc_skb_received);
c247f053 WB	174	serr = SKB_EXT_ERR(skb);
c247f053 WB	175	if (!skb->len && serr->ee.ee_origin == SO_EE_ORIGIN_TIMESTAMPING) {
49ca0d8b	176	_leave("UDP empty message");
2ca4f6ca	177	rcu_read_unlock();
987db9f7	178	rxrpc_free_skb(skb, rxrpc_skb_freed);
49ca0d8b WB	179	return;
49ca0d8b WB	180	}
17926a79	181
494337c9	182	peer = rxrpc_lookup_peer_icmp_rcu(local, skb, &srx);
be6e6707 DH	183	if (peer && !rxrpc_get_peer_maybe(peer))
	184	peer = NULL;
	185	if (!peer) {
	186	rcu_read_unlock();
987db9f7	187	rxrpc_free_skb(skb, rxrpc_skb_freed);
17926a79 DH	188	_leave(" [no peer]");
	189	return;
	190	}
	191
494337c9 DH	192	trace_rxrpc_rx_icmp(peer, &serr->ee, &srx);
494337c9 DH	193
1a70c05b DH	194	if ((serr->ee.ee_origin == SO_EE_ORIGIN_ICMP &&
	195	serr->ee.ee_type == ICMP_DEST_UNREACH &&
	196	serr->ee.ee_code == ICMP_FRAG_NEEDED)) {
	197	rxrpc_adjust_mtu(peer, serr);
f66d7490	198	rcu_read_unlock();
987db9f7	199	rxrpc_free_skb(skb, rxrpc_skb_freed);
f66d7490 DH	200	rxrpc_put_peer(peer);
	201	_leave(" [MTU update]");
	202	return;
17926a79 DH	203	}
17926a79 DH	204
f66d7490	205	rxrpc_store_error(peer, serr);
be6e6707	206	rcu_read_unlock();
987db9f7	207	rxrpc_free_skb(skb, rxrpc_skb_freed);
1890fea7	208	rxrpc_put_peer(peer);
17926a79	209
17926a79 DH	210	_leave("");
	211	}
	212
	213	/*
f66d7490	214	* Map an error report to error codes on the peer record.
17926a79	215	*/
f66d7490 DH	216	static void rxrpc_store_error(struct rxrpc_peer *peer,
f66d7490 DH	217	struct sock_exterr_skb *serr)
17926a79	218	{
f3344303	219	enum rxrpc_call_completion compl = RXRPC_CALL_NETWORK_ERROR;
17926a79	220	struct sock_extended_err *ee;
c9d10c49	221	int err;
17926a79 DH	222
	223	_enter("");
	224
17926a79 DH	225	ee = &serr->ee;
17926a79 DH	226
17926a79 DH	227	err = ee->ee_errno;
	228
	229	switch (ee->ee_origin) {
	230	case SO_EE_ORIGIN_ICMP:
17926a79 DH	231	switch (ee->ee_type) {
	232	case ICMP_DEST_UNREACH:
	233	switch (ee->ee_code) {
	234	case ICMP_NET_UNREACH:
	235	_net("Rx Received ICMP Network Unreachable");
17926a79 DH	236	break;
	237	case ICMP_HOST_UNREACH:
	238	_net("Rx Received ICMP Host Unreachable");
17926a79 DH	239	break;
	240	case ICMP_PORT_UNREACH:
	241	_net("Rx Received ICMP Port Unreachable");
17926a79 DH	242	break;
	243	case ICMP_NET_UNKNOWN:
	244	_net("Rx Received ICMP Unknown Network");
17926a79 DH	245	break;
	246	case ICMP_HOST_UNKNOWN:
	247	_net("Rx Received ICMP Unknown Host");
17926a79 DH	248	break;
	249	default:
	250	_net("Rx Received ICMP DestUnreach code=%u",
	251	ee->ee_code);
	252	break;
	253	}
	254	break;
	255
	256	case ICMP_TIME_EXCEEDED:
	257	_net("Rx Received ICMP TTL Exceeded");
	258	break;
	259
	260	default:
	261	_proto("Rx Received ICMP error { type=%u code=%u }",
	262	ee->ee_type, ee->ee_code);
	263	break;
	264	}
	265	break;
	266
f66d7490	267	case SO_EE_ORIGIN_NONE:
17926a79	268	case SO_EE_ORIGIN_LOCAL:
fe77d5fc	269	_proto("Rx Received local error { error=%d }", err);
f3344303	270	compl = RXRPC_CALL_LOCAL_ERROR;
17926a79 DH	271	break;
17926a79 DH	272
17926a79 DH	273	case SO_EE_ORIGIN_ICMP6:
17926a79 DH	274	default:
fe77d5fc	275	_proto("Rx Received error report { orig=%u }", ee->ee_origin);
17926a79 DH	276	break;
	277	}
	278
f3344303	279	rxrpc_distribute_error(peer, err, compl);
f66d7490 DH	280	}
	281
	282	/*
f3344303	283	* Distribute an error that occurred on a peer.
f66d7490	284	*/
f3344303 DH	285	static void rxrpc_distribute_error(struct rxrpc_peer *peer, int error,
f3344303 DH	286	enum rxrpc_call_completion compl)
f66d7490	287	{
f66d7490	288	struct rxrpc_call *call;
17926a79	289
f3344303	290	hlist_for_each_entry_rcu(call, &peer->error_targets, error_link) {
e34d4234	291	rxrpc_see_call(call);
f3344303 DH	292	if (call->state < RXRPC_CALL_COMPLETE &&
f3344303 DH	293	rxrpc_set_call_completion(call, compl, 0, -error))
248f219c	294	rxrpc_notify_socket(call);
17926a79	295	}
17926a79	296	}
cf1a6474 DH	297
	298	/*
	299	* Add RTT information to cache. This is called in softirq mode and has
	300	* exclusive access to the peer RTT data.
	301	*/
	302	void rxrpc_peer_add_rtt(struct rxrpc_call *call, enum rxrpc_rtt_rx_trace why,
	303	rxrpc_serial_t send_serial, rxrpc_serial_t resp_serial,
	304	ktime_t send_time, ktime_t resp_time)
	305	{
	306	struct rxrpc_peer *peer = call->peer;
	307	s64 rtt;
	308	u64 sum = peer->rtt_sum, avg;
	309	u8 cursor = peer->rtt_cursor, usage = peer->rtt_usage;
	310
	311	rtt = ktime_to_ns(ktime_sub(resp_time, send_time));
	312	if (rtt < 0)
	313	return;
	314
c1e15b49 DH	315	spin_lock(&peer->rtt_input_lock);
c1e15b49 DH	316
cf1a6474 DH	317	/* Replace the oldest datum in the RTT buffer */
	318	sum -= peer->rtt_cache[cursor];
	319	sum += rtt;
	320	peer->rtt_cache[cursor] = rtt;
	321	peer->rtt_cursor = (cursor + 1) & (RXRPC_RTT_CACHE_SIZE - 1);
	322	peer->rtt_sum = sum;
	323	if (usage < RXRPC_RTT_CACHE_SIZE) {
	324	usage++;
	325	peer->rtt_usage = usage;
	326	}
	327
c1e15b49 DH	328	spin_unlock(&peer->rtt_input_lock);
c1e15b49 DH	329
cf1a6474 DH	330	/* Now recalculate the average */
	331	if (usage == RXRPC_RTT_CACHE_SIZE) {
	332	avg = sum / RXRPC_RTT_CACHE_SIZE;
	333	} else {
	334	avg = sum;
	335	do_div(avg, usage);
	336	}
	337
c1e15b49	338	/* Don't need to update this under lock */
cf1a6474 DH	339	peer->rtt = avg;
	340	trace_rxrpc_rtt_rx(call, why, send_serial, resp_serial, rtt,
	341	usage, avg);
	342	}
ace45bec DH	343
ace45bec DH	344	/*
330bdcfa	345	* Perform keep-alive pings.
ace45bec	346	*/
330bdcfa DH	347	static void rxrpc_peer_keepalive_dispatch(struct rxrpc_net *rxnet,
	348	struct list_head *collector,
	349	time64_t base,
	350	u8 cursor)
ace45bec	351	{
ace45bec	352	struct rxrpc_peer *peer;
330bdcfa DH	353	const u8 mask = ARRAY_SIZE(rxnet->peer_keepalive) - 1;
	354	time64_t keepalive_at;
	355	int slot;
ace45bec	356
330bdcfa	357	spin_lock_bh(&rxnet->peer_hash_lock);
ace45bec	358
330bdcfa DH	359	while (!list_empty(collector)) {
	360	peer = list_entry(collector->next,
	361	struct rxrpc_peer, keepalive_link);
ace45bec	362
330bdcfa DH	363	list_del_init(&peer->keepalive_link);
	364	if (!rxrpc_get_peer_maybe(peer))
	365	continue;
ace45bec	366
ace45bec	367	spin_unlock_bh(&rxnet->peer_hash_lock);
ace45bec	368
330bdcfa DH	369	keepalive_at = peer->last_tx_at + RXRPC_KEEPALIVE_TIME;
	370	slot = keepalive_at - base;
	371	_debug("%02x peer %u t=%d {%pISp}",
	372	cursor, peer->debug_id, slot, &peer->srx.transport);
	373
	374	if (keepalive_at <= base \|\|
	375	keepalive_at > base + RXRPC_KEEPALIVE_TIME) {
	376	rxrpc_send_keepalive(peer);
	377	slot = RXRPC_KEEPALIVE_TIME;
ace45bec DH	378	}
ace45bec DH	379
330bdcfa DH	380	/* A transmission to this peer occurred since last we examined
	381	* it so put it into the appropriate future bucket.
	382	*/
	383	slot += cursor;
	384	slot &= mask;
	385	spin_lock_bh(&rxnet->peer_hash_lock);
	386	list_add_tail(&peer->keepalive_link,
	387	&rxnet->peer_keepalive[slot & mask]);
60034d3d	388	rxrpc_put_peer_locked(peer);
ace45bec DH	389	}
ace45bec DH	390
ace45bec	391	spin_unlock_bh(&rxnet->peer_hash_lock);
330bdcfa	392	}
ace45bec	393
330bdcfa DH	394	/*
	395	* Perform keep-alive pings with VERSION packets to keep any NAT alive.
	396	*/
	397	void rxrpc_peer_keepalive_worker(struct work_struct *work)
	398	{
	399	struct rxrpc_net *rxnet =
	400	container_of(work, struct rxrpc_net, peer_keepalive_work);
	401	const u8 mask = ARRAY_SIZE(rxnet->peer_keepalive) - 1;
	402	time64_t base, now, delay;
	403	u8 cursor, stop;
	404	LIST_HEAD(collector);
ace45bec	405
330bdcfa DH	406	now = ktime_get_seconds();
	407	base = rxnet->peer_keepalive_base;
	408	cursor = rxnet->peer_keepalive_cursor;
	409	_enter("%lld,%u", base - now, cursor);
ace45bec	410
330bdcfa DH	411	if (!rxnet->live)
330bdcfa DH	412	return;
ace45bec	413
330bdcfa DH	414	/* Remove to a temporary list all the peers that are currently lodged
	415	* in expired buckets plus all new peers.
	416	*
	417	* Everything in the bucket at the cursor is processed this
	418	* second; the bucket at cursor + 1 goes at now + 1s and so
	419	* on...
ace45bec	420	*/
ace45bec	421	spin_lock_bh(&rxnet->peer_hash_lock);
330bdcfa DH	422	list_splice_init(&rxnet->peer_keepalive_new, &collector);
	423
	424	stop = cursor + ARRAY_SIZE(rxnet->peer_keepalive);
	425	while (base <= now && (s8)(cursor - stop) < 0) {
	426	list_splice_tail_init(&rxnet->peer_keepalive[cursor & mask],
	427	&collector);
	428	base++;
	429	cursor++;
	430	}
ace45bec	431
330bdcfa DH	432	base = now;
330bdcfa DH	433	spin_unlock_bh(&rxnet->peer_hash_lock);
ace45bec	434
ace45bec DH	435	rxnet->peer_keepalive_base = base;
ace45bec DH	436	rxnet->peer_keepalive_cursor = cursor;
330bdcfa DH	437	rxrpc_peer_keepalive_dispatch(rxnet, &collector, base, cursor);
	438	ASSERT(list_empty(&collector));
	439
	440	/* Schedule the timer for the next occupied timeslot. */
	441	cursor = rxnet->peer_keepalive_cursor;
	442	stop = cursor + RXRPC_KEEPALIVE_TIME - 1;
	443	for (; (s8)(cursor - stop) < 0; cursor++) {
	444	if (!list_empty(&rxnet->peer_keepalive[cursor & mask]))
	445	break;
	446	base++;
	447	}
	448
	449	now = ktime_get_seconds();
	450	delay = base - now;
	451	if (delay < 1)
	452	delay = 1;
	453	delay *= HZ;
	454	if (rxnet->live)
	455	timer_reduce(&rxnet->peer_keepalive_timer, jiffies + delay);
	456
ace45bec DH	457	_leave("");
ace45bec DH	458	}