path: root/net/mctp/route.c

// SPDX-License-Identifier: GPL-2.0
/*
 * Management Component Transport Protocol (MCTP) - routing
 * implementation.
 *
 * This is currently based on a simple routing table, with no dst cache. The
 * number of routes should stay fairly small, so the lookup cost is small.
 *
 * Copyright (c) 2021 Code Construct
 * Copyright (c) 2021 Google
 */

#include <linux/idr.h>
#include <linux/kconfig.h>
#include <linux/mctp.h>
#include <linux/netdevice.h>
#include <linux/rtnetlink.h>
#include <linux/skbuff.h>

#include <uapi/linux/if_arp.h>

#include <net/mctp.h>
#include <net/mctpdevice.h>
#include <net/netlink.h>
#include <net/sock.h>

#include <trace/events/mctp.h>

static const unsigned int mctp_message_maxlen = 64 * 1024;
static const unsigned long mctp_key_lifetime = 6 * CONFIG_HZ;

static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev);

/* route output callbacks */
static int mctp_route_discard(struct mctp_route *route, struct sk_buff *skb)
{
	kfree_skb(skb);
	return 0;
}

static struct mctp_sock *mctp_lookup_bind(struct net *net, struct sk_buff *skb)
{
	struct mctp_skb_cb *cb = mctp_cb(skb);
	struct mctp_hdr *mh;
	struct sock *sk;
	u8 type;

	WARN_ON(!rcu_read_lock_held());

	/* TODO: look up in skb->cb? */
	mh = mctp_hdr(skb);

	if (!skb_headlen(skb))
		return NULL;

	type = (*(u8 *)skb->data) & 0x7f;

	sk_for_each_rcu(sk, &net->mctp.binds) {
		struct mctp_sock *msk = container_of(sk, struct mctp_sock, sk);

		if (msk->bind_net != MCTP_NET_ANY && msk->bind_net != cb->net)
			continue;

		if (msk->bind_type != type)
			continue;

		if (!mctp_address_matches(msk->bind_addr, mh->dest))
			continue;

		return msk;
	}

	return NULL;
}

/* A note on the key allocations.
 *
 * struct net->mctp.keys contains our set of currently-allocated keys for
 * MCTP tag management. The lookup tuple for these is the peer EID,
 * local EID and MCTP tag.
 *
 * In some cases, the peer EID may be MCTP_EID_ANY: for example, when a
 * broadcast message is sent, we may receive responses from any peer EID.
 * Because the broadcast dest address is equivalent to ANY, we create
 * a key with (local = local-eid, peer = ANY). This allows a match on the
 * incoming broadcast responses from any peer.
 *
 * We perform lookups when packets are received, and when tags are allocated
 * in two scenarios:
 *
 *  - when a packet is sent, with a locally-owned tag: we need to find an
 *    unused tag value for the (local, peer) EID pair.
 *
 *  - when a tag is manually allocated: we need to find an unused tag value
 *    for the peer EID, but don't have a specific local EID at that stage.
 *
 * in the latter case, on successful allocation, we end up with a tag with
 * (local = ANY, peer = peer-eid).
 *
 * So, the key set allows both a local EID of ANY, as well as a peer EID of
 * ANY in the lookup tuple. Both may be ANY if we prealloc for a broadcast.
 * The matching (in mctp_key_match()) during lookup allows the match value to
 * be ANY in either the dest or source addresses.
 *
 * When allocating (+ inserting) a tag, we need to check for conflicts amongst
 * the existing tag set. This requires macthing either exactly on the local
 * and peer addresses, or either being ANY.
 */

static bool mctp_key_match(struct mctp_sk_key *key, unsigned int net,
			   mctp_eid_t local, mctp_eid_t peer, u8 tag)
{
	if (key->net != net)
		return false;

	if (!mctp_address_matches(key->local_addr, local))
		return false;

	if (!mctp_address_matches(key->peer_addr, peer))
		return false;

	if (key->tag != tag)
		return false;

	return true;
}

/* returns a key (with key->lock held, and refcounted), or NULL if no such
 * key exists.
 */
static struct mctp_sk_key *mctp_lookup_key(struct net *net, struct sk_buff *skb,
					   unsigned int netid, mctp_eid_t peer,
					   unsigned long *irqflags)
	__acquires(&key->lock)
{
	struct mctp_sk_key *key, *ret;
	unsigned long flags;
	struct mctp_hdr *mh;
	u8 tag;

	mh = mctp_hdr(skb);
	tag = mh->flags_seq_tag & (MCTP_HDR_TAG_MASK | MCTP_HDR_FLAG_TO);

	ret = NULL;
	spin_lock_irqsave(&net->mctp.keys_lock, flags);

	hlist_for_each_entry(key, &net->mctp.keys, hlist) {
		if (!mctp_key_match(key, netid, mh->dest, peer, tag))
			continue;

		spin_lock(&key->lock);
		if (key->valid) {
			refcount_inc(&key->refs);
			ret = key;
			break;
		}
		spin_unlock(&key->lock);
	}

	if (ret) {
		spin_unlock(&net->mctp.keys_lock);
		*irqflags = flags;
	} else {
		spin_unlock_irqrestore(&net->mctp.keys_lock, flags);
	}

	return ret;
}

static struct mctp_sk_key *mctp_key_alloc(struct mctp_sock *msk,
					  unsigned int net,
					  mctp_eid_t local, mctp_eid_t peer,
					  u8 tag, gfp_t gfp)
{
	struct mctp_sk_key *key;

	key = kzalloc(sizeof(*key), gfp);
	if (!key)
		return NULL;

	key->net = net;
	key->peer_addr = peer;
	key->local_addr = local;
	key->tag = tag;
	key->sk = &msk->sk;
	key->valid = true;
	spin_lock_init(&key->lock);
	refcount_set(&key->refs, 1);
	sock_hold(key->sk);

	return key;
}

void mctp_key_unref(struct mctp_sk_key *key)
{
	unsigned long flags;

	if (!refcount_dec_and_test(&key->refs))
		return;

	/* even though no refs exist here, the lock allows us to stay
	 * consistent with the locking requirement of mctp_dev_release_key
	 */
	spin_lock_irqsave(&key->lock, flags);
	mctp_dev_release_key(key->dev, key);
	spin_unlock_irqrestore(&key->lock, flags);

	sock_put(key->sk);
	kfree(key);
}

static int mctp_key_add(struct mctp_sk_key *key, struct mctp_sock *msk)
{
	struct net *net = sock_net(&msk->sk);
	struct mctp_sk_key *tmp;
	unsigned long flags;
	int rc = 0;

	spin_lock_irqsave(&net->mctp.keys_lock, flags);

	if (sock_flag(&msk->sk, SOCK_DEAD)) {
		rc = -EINVAL;
		goto out_unlock;
	}

	hlist_for_each_entry(tmp, &net->mctp.keys, hlist) {
		if (mctp_key_match(tmp, key->net, key->local_addr,
				   key->peer_addr, key->tag)) {
			spin_lock(&tmp->lock);
			if (tmp->valid)
				rc = -EEXIST;
			spin_unlock(&tmp->lock);
			if (rc)
				break;
		}
	}

	if (!rc) {
		refcount_inc(&key->refs);
		key->expiry = jiffies + mctp_key_lifetime;
		timer_reduce(&msk->key_expiry, key->expiry);

		hlist_add_head(&key->hlist, &net->mctp.keys);
		hlist_add_head(&key->sklist, &msk->keys);
	}

out_unlock:
	spin_unlock_irqrestore(&net->mctp.keys_lock, flags);

	return rc;
}

/* Helper for mctp_route_input().
 * We're done with the key; unlock and unref the key.
 * For the usual case of automatic expiry we remove the key from lists.
 * In the case that manual allocation is set on a key we release the lock
 * and local ref, reset reassembly, but don't remove from lists.
 */
static void __mctp_key_done_in(struct mctp_sk_key *key, struct net *net,
			       unsigned long flags, unsigned long reason)
__releases(&key->lock)
{
	struct sk_buff *skb;

	trace_mctp_key_release(key, reason);
	skb = key->reasm_head;
	key->reasm_head = NULL;

	if (!key->manual_alloc) {
		key->reasm_dead = true;
		key->valid = false;
		mctp_dev_release_key(key->dev, key);
	}
	spin_unlock_irqrestore(&key->lock, flags);

	if (!key->manual_alloc) {
		spin_lock_irqsave(&net->mctp.keys_lock, flags);
		if (!hlist_unhashed(&key->hlist)) {
			hlist_del_init(&key->hlist);
			hlist_del_init(&key->sklist);
			mctp_key_unref(key);
		}
		spin_unlock_irqrestore(&net->mctp.keys_lock, flags);
	}

	/* and one for the local reference */
	mctp_key_unref(key);

	kfree_skb(skb);
}

#ifdef CONFIG_MCTP_FLOWS
static void mct