Merge branch 'MCTP-flow-support'

Jeremy Kerr says:

====================
MCTP flow support

For certain MCTP transport bindings, the binding driver will need to be
aware of request/response pairing. For example, the i2c binding may need
to set multiplexer configuration when expecting a response from a device
behind a mux.

This series implements a mechanism for informing the driver about these
flows, so it can implement transport-specific behaviour when a flow is
in progress (ie, a response is expected, and/or we time-out on that
expectation). We use a skb extension to notify the driver about the
presence of a flow, and a new dev->ops callback to notify about a flow's
destruction.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/skbuff.h

@@ -4243,6 +4243,9 @@ enum skb_ext_id {
 #endif
 #if IS_ENABLED(CONFIG_MPTCP)
 	SKB_EXT_MPTCP,
+#endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+	SKB_EXT_MCTP,
 #endif
 	SKB_EXT_NUM, /* must be last */
 };

include/net/mctp.h

@@ -152,6 +152,12 @@ struct mctp_sk_key {
 
 	/* expiry timeout; valid (above) cleared on expiry */
 	unsigned long	expiry;
+
+	/* free to use for device flow state tracking. Initialised to
+	 * zero on initial key creation
+	 */
+	unsigned long	dev_flow_state;
+	struct mctp_dev	*dev;
 };
 
 struct mctp_skb_cb {
@@ -189,6 +195,13 @@ static inline struct mctp_skb_cb *mctp_cb(struct sk_buff *skb)
 	return (void *)(skb->cb);
 }
 
+/* If CONFIG_MCTP_FLOWS, we may add one of these as a SKB extension,
+ * indicating the flow to the device driver.
+ */
+struct mctp_flow {
+	struct mctp_sk_key *key;
+};
+
 /* Route definition.
  *
  * These are held in the pernet->mctp.routes list, with RCU protection for

include/net/mctpdevice.h

@@ -14,6 +14,8 @@
 #include <linux/types.h>
 #include <linux/refcount.h>
 
+struct mctp_sk_key;
+
 struct mctp_dev {
 	struct net_device	*dev;
 
@@ -21,6 +23,8 @@ struct mctp_dev {
 
 	unsigned int		net;
 
+	const struct mctp_netdev_ops *ops;
+
 	/* Only modified under RTNL. Reads have addrs_lock held */
 	u8			*addrs;
 	size_t			num_addrs;
@@ -29,12 +33,24 @@ struct mctp_dev {
 	struct rcu_head		rcu;
 };
 
+struct mctp_netdev_ops {
+	void			(*release_flow)(struct mctp_dev *dev,
+						struct mctp_sk_key *key);
+};
+
 #define MCTP_INITIAL_DEFAULT_NET	1
 
 struct mctp_dev *mctp_dev_get_rtnl(const struct net_device *dev);
 struct mctp_dev *__mctp_dev_get(const struct net_device *dev);
 
+int mctp_register_netdev(struct net_device *dev,
+			 const struct mctp_netdev_ops *ops);
+void mctp_unregister_netdev(struct net_device *dev);
+
 void mctp_dev_hold(struct mctp_dev *mdev);
 void mctp_dev_put(struct mctp_dev *mdev);
 
+void mctp_dev_set_key(struct mctp_dev *dev, struct mctp_sk_key *key);
+void mctp_dev_release_key(struct mctp_dev *dev, struct mctp_sk_key *key);
+
 #endif /* __NET_MCTPDEVICE_H */

net/core/skbuff.c

@@ -70,6 +70,7 @@
 #include <net/xfrm.h>
 #include <net/mpls.h>
 #include <net/mptcp.h>
+#include <net/mctp.h>
 #include <net/page_pool.h>
 
 #include <linux/uaccess.h>
@@ -4440,6 +4441,9 @@ static const u8 skb_ext_type_len[] = {
 #if IS_ENABLED(CONFIG_MPTCP)
 	[SKB_EXT_MPTCP] = SKB_EXT_CHUNKSIZEOF(struct mptcp_ext),
 #endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+	[SKB_EXT_MCTP] = SKB_EXT_CHUNKSIZEOF(struct mctp_flow),
+#endif
 };
 
 static __always_inline unsigned int skb_ext_total_length(void)
@@ -4456,6 +4460,9 @@ static __always_inline unsigned int skb_ext_total_length(void)
 #endif
 #if IS_ENABLED(CONFIG_MPTCP)
 		skb_ext_type_len[SKB_EXT_MPTCP] +
+#endif
+#if IS_ENABLED(CONFIG_MCTP_FLOWS)
+		skb_ext_type_len[SKB_EXT_MCTP] +
 #endif
 		0;
 }
@@ -6529,6 +6536,14 @@ static void skb_ext_put_sp(struct sec_path *sp)
 }
 #endif
 
+#ifdef CONFIG_MCTP_FLOWS
+static void skb_ext_put_mctp(struct mctp_flow *flow)
+{
+	if (flow->key)
+		mctp_key_unref(flow->key);
+}
+#endif
+
 void __skb_ext_del(struct sk_buff *skb, enum skb_ext_id id)
 {
 	struct skb_ext *ext = skb->extensions;
@@ -6564,6 +6579,10 @@ void __skb_ext_put(struct skb_ext *ext)
 	if (__skb_ext_exist(ext, SKB_EXT_SEC_PATH))
 		skb_ext_put_sp(skb_ext_get_ptr(ext, SKB_EXT_SEC_PATH));
 #endif
+#ifdef CONFIG_MCTP_FLOWS
+	if (__skb_ext_exist(ext, SKB_EXT_MCTP))
+		skb_ext_put_mctp(skb_ext_get_ptr(ext, SKB_EXT_MCTP));
+#endif
 
 	kmem_cache_free(skbuff_ext_cache, ext);
 }

net/mctp/Kconfig

 
 menuconfig MCTP
 	depends on NET
-	tristate "MCTP core protocol support"
+	bool "MCTP core protocol support"
 	help
 	  Management Component Transport Protocol (MCTP) is an in-system
 	  protocol for communicating between management controllers and
@@ -16,3 +16,8 @@ config MCTP_TEST
         bool "MCTP core tests" if !KUNIT_ALL_TESTS
         depends on MCTP=y && KUNIT=y
         default KUNIT_ALL_TESTS
+
+config MCTP_FLOWS
+	bool
+	depends on MCTP
+	select SKB_EXTENSIONS

net/mctp/device.c

@@ -260,6 +260,24 @@ void mctp_dev_put(struct mctp_dev *mdev)
 	}
 }
 
+void mctp_dev_release_key(struct mctp_dev *dev, struct mctp_sk_key *key)
+	__must_hold(&key->lock)
+{
+	if (!dev)
+		return;
+	if (dev->ops && dev->ops->release_flow)
+		dev->ops->release_flow(dev, key);
+	key->dev = NULL;
+	mctp_dev_put(dev);
+}
+
+void mctp_dev_set_key(struct mctp_dev *dev, struct mctp_sk_key *key)
+	__must_hold(&key->lock)
+{
+	mctp_dev_hold(dev);
+	key->dev = dev;
+}
+
 static struct mctp_dev *mctp_add_dev(struct net_device *dev)
 {
 	struct mctp_dev *mdev;
@@ -414,6 +432,39 @@ static int mctp_dev_notify(struct notifier_block *this, unsigned long event,
 	return NOTIFY_OK;
 }
 
+static int mctp_register_netdevice(struct net_device *dev,
+				   const struct mctp_netdev_ops *ops)
+{
+	struct mctp_dev *mdev;
+
+	mdev = mctp_add_dev(dev);
+	if (IS_ERR(mdev))
+		return PTR_ERR(mdev);
+
+	mdev->ops = ops;
+
+	return register_netdevice(dev);
+}
+
+int mctp_register_netdev(struct net_device *dev,
+			 const struct mctp_netdev_ops *ops)
+{
+	int rc;
+
+	rtnl_lock();
+	rc = mctp_register_netdevice(dev, ops);
+	rtnl_unlock();
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(mctp_register_netdev);
+
+void mctp_unregister_netdev(struct net_device *dev)
+{
+	unregister_netdev(dev);
+}
+EXPORT_SYMBOL_GPL(mctp_unregister_netdev);
+
 static struct rtnl_af_ops mctp_af_ops = {
 	.family = AF_MCTP,
 	.fill_link_af = mctp_fill_link_af,

net/mctp/route.c

@@ -29,6 +29,8 @@
 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)
 {
@@ -152,8 +154,19 @@ static struct mctp_sk_key *mctp_key_alloc(struct mctp_sock *msk,
 
 void mctp_key_unref(struct mctp_sk_key *key)
 {
-	if (refcount_dec_and_test(&key->refs))
-		kfree(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);
+
+	kfree(key);
 }
 
 static int mctp_key_add(struct mctp_sk_key *key, struct mctp_sock *msk)
@@ -204,6 +217,7 @@ static void __mctp_key_unlock_drop(struct mctp_sk_key *key, struct net *net,
 	key->reasm_head = NULL;
 	key->reasm_dead = true;
 	key->valid = false;
+	mctp_dev_release_key(key->dev, key);
 	spin_unlock_irqrestore(&key->lock, flags);
 
 	spin_lock_irqsave(&net->mctp.keys_lock, flags);
@@ -222,6 +236,40 @@ static void __mctp_key_unlock_drop(struct mctp_sk_key *key, struct net *net,
 
 }
 
+#ifdef CONFIG_MCTP_FLOWS
+static void mctp_skb_set_flow(struct sk_buff *skb, struct mctp_sk_key *key)
+{
+	struct mctp_flow *flow;
+
+	flow = skb_ext_add(skb, SKB_EXT_MCTP);
+	if (!flow)
+		return;
+
+	refcount_inc(&key->refs);
+	flow->key = key;
+}
+
+static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev)
+{
+	struct mctp_sk_key *key;
+	struct mctp_flow *flow;
+
+	flow = skb_ext_find(skb, SKB_EXT_MCTP);
+	if (!flow)
+		return;
+
+	key = flow->key;
+
+	if (WARN_ON(key->dev && key->dev != dev))
+		return;
+
+	mctp_dev_set_key(dev, key);
+}
+#else
+static void mctp_skb_set_flow(struct sk_buff *skb, struct mctp_sk_key *key) {}
+static void mctp_flow_prepare_output(struct sk_buff *skb, struct mctp_dev *dev) {}
+#endif
+
 static int mctp_frag_queue(struct mctp_sk_key *key, struct sk_buff *skb)
 {
 	struct mctp_hdr *hdr = mctp_hdr(skb);
@@ -465,6 +513,8 @@ static int mctp_route_output(struct mctp_route *route, struct sk_buff *skb)
 		return -EHOSTUNREACH;
 	}
 
+	mctp_flow_prepare_output(skb, route->dev);
+
 	rc = dev_queue_xmit(skb);
 	if (rc)
 		rc = net_xmit_errno(rc);
@@ -532,14 +582,14 @@ static void mctp_reserve_tag(struct net *net, struct mctp_sk_key *key,
 /* Allocate a locally-owned tag value for (saddr, daddr), and reserve
  * it for the socket msk
  */
-static int mctp_alloc_local_tag(struct mctp_sock *msk,
-				mctp_eid_t saddr, mctp_eid_t daddr, u8 *tagp)
+static struct mctp_sk_key *mctp_alloc_local_tag(struct mctp_sock *msk,
+						mctp_eid_t saddr,
+						mctp_eid_t daddr, u8 *tagp)
 {
 	struct net *net = sock_net(&msk->sk);
 	struct netns_mctp *mns = &net->mctp;
 	struct mctp_sk_key *key, *tmp;
 	unsigned long flags;
-	int rc = -EAGAIN;
 	u8 tagbits;
 
 	/* for NULL destination EIDs, we may get a response from any peer */
@@ -549,7 +599,7 @@ static int mctp_alloc_local_tag(struct mctp_sock *msk,
 	/* be optimistic, alloc now */
 	key = mctp_key_alloc(msk, saddr, daddr, 0, GFP_KERNEL);
 	if (!key)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	/* 8 possible tag values */
 	tagbits = 0xff;
@@ -591,18 +641,16 @@ static int mctp_alloc_local_tag(struct mctp_sock *msk,
 		trace_mctp_key_acquire(key);
 
 		*tagp = key->tag;
-		/* done with the key in this scope */
-		mctp_key_unref(key);
-		key = NULL;
-		rc = 0;
 	}
 
 	spin_unlock_irqrestore(&mns->keys_lock, flags);
 
-	if (!tagbits)
+	if (!tagbits) {
 		kfree(key);
+		return ERR_PTR(-EBUSY);
+	}
 
-	return rc;
+	return key;
 }
 
 /* routing lookups */
@@ -740,6 +788,7 @@ int mctp_local_output(struct sock *sk, struct mctp_route *rt,
 	struct mctp_sock *msk = container_of(sk, struct mctp_sock, sk);
 	struct mctp_skb_cb *cb = mctp_cb(skb);
 	struct mctp_route tmp_rt;
+	struct mctp_sk_key *key;
 	struct net_device *dev;
 	struct mctp_hdr *hdr;
 	unsigned long flags;
@@ -799,11 +848,17 @@ int mctp_local_output(struct sock *sk, struct mctp_route *rt,
 		goto out_release;
 
 	if (req_tag & MCTP_HDR_FLAG_TO) {
-		rc = mctp_alloc_local_tag(msk, saddr, daddr, &tag);
-		if (rc)
+		key = mctp_alloc_local_tag(msk, saddr, daddr, &tag);
+		if (IS_ERR(key)) {
+			rc = PTR_ERR(key);
 			goto out_release;
+		}
+		mctp_skb_set_flow(skb, key);
+		/* done with the key in this scope */
+		mctp_key_unref(key);
 		tag |= MCTP_HDR_FLAG_TO;
 	} else {
+		key = NULL;
 		tag = req_tag;
 	}