Merge tag 'net-5.13-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Networking fixes, including fixes from bpf, wireless, netfilter and
  wireguard trees.

  The bpf vs lockdown+audit fix is the most notable.

  Things haven't slowed down just yet, both in terms of regressions in
  current release and largish fixes for older code, but we usually see a
  slowdown only after -rc5.

  Current release - regressions:

   - virtio-net: fix page faults and crashes when XDP is enabled

   - mlx5e: fix HW timestamping with CQE compression, and make sure they
     are only allowed to coexist with capable devices

   - stmmac:
      - fix kernel panic due to NULL pointer dereference of
        mdio_bus_data
      - fix double clk unprepare when no PHY device is connected

  Current release - new code bugs:

   - mt76: a few fixes for the recent MT7921 devices and runtime power
     management

  Previous releases - regressions:

   - ice:
      - track AF_XDP ZC enabled queues in bitmap to fix copy mode Tx
      - fix allowing VF to request more/less queues via virtchnl
      - correct supported and advertised autoneg by using PHY
        capabilities
      - allow all LLDP packets from PF to Tx

   - kbuild: quote OBJCOPY var to avoid a pahole call break the build

  Previous releases - always broken:

   - bpf, lockdown, audit: fix buggy SELinux lockdown permission checks

   - mt76: address the recent FragAttack vulnerabilities not covered by
     generic fixes

   - ipv6: fix KASAN: slab-out-of-bounds Read in
     fib6_nh_flush_exceptions

   - Bluetooth:
      - fix the erroneous flush_work() order, to avoid double free
      - use correct lock to prevent UAF of hdev object

   - nfc: fix NULL ptr dereference in llcp_sock_getname() after failed
     connect

   - ieee802154: multiple fixes to error checking and return values

   - igb: fix XDP with PTP enabled

   - intel: add correct exception tracing for XDP

   - tls: fix use-after-free when TLS offload device goes down and back
     up

   - ipvs: ignore IP_VS_SVC_F_HASHED flag when adding service

   - netfilter: nft_ct: skip expectations for confirmed conntrack

   - mptcp: fix falling back to TCP in presence of out of order packets
     early in connection lifetime

   - wireguard: switch from O(n) to a O(1) algorithm for maintaining
     peers, fixing stalls and a large memory leak in the process

  Misc:

   - devlink: correct VIRTUAL port to not have phys_port attributes

   - Bluetooth: fix VIRTIO_ID_BT assigned number

   - net: return the correct errno code ENOBUF -> ENOMEM

   - wireguard:
      - peer: allocate in kmem_cache saving 25% on peer memory
      - do not use -O3"

* tag 'net-5.13-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (91 commits)
  cxgb4: avoid link re-train during TC-MQPRIO configuration
  sch_htb: fix refcount leak in htb_parent_to_leaf_offload
  wireguard: allowedips: free empty intermediate nodes when removing single node
  wireguard: allowedips: allocate nodes in kmem_cache
  wireguard: allowedips: remove nodes in O(1)
  wireguard: allowedips: initialize list head in selftest
  wireguard: peer: allocate in kmem_cache
  wireguard: use synchronize_net rather than synchronize_rcu
  wireguard: do not use -O3
  wireguard: selftests: make sure rp_filter is disabled on vethc
  wireguard: selftests: remove old conntrack kconfig value
  virtchnl: Add missing padding to virtchnl_proto_hdrs
  ice: Allow all LLDP packets from PF to Tx
  ice: report supported and advertised autoneg using PHY capabilities
  ice: handle the VF VSI rebuild failure
  ice: Fix VFR issues for AVF drivers that expect ATQLEN cleared
  ice: Fix allowing VF to request more/less queues via virtchnl
  virtio-net: fix for skb_over_panic inside big mode
  ipv6: Fix KASAN: slab-out-of-bounds Read in fib6_nh_flush_exceptions
  fib: Return the correct errno code
  ...

1 files changed, 99 insertions, 90 deletions

diff --git a/drivers/net/wireguard/allowedips.c b/drivers/net/wireguard/allowedips.c
index 3725e9cd85f4..b7197e80f226 100644
--- a/drivers/net/wireguard/allowedips.c
+++ b/drivers/net/wireguard/allowedips.c
@@ -6,6 +6,8 @@
 #include "allowedips.h"
 #include "peer.h"
 
+static struct kmem_cache *node_cache;
+
 static void swap_endian(u8 *dst, const u8 *src, u8 bits)
 {
 	if (bits == 32) {
@@ -28,8 +30,11 @@ static void copy_and_assign_cidr(struct allowedips_node *node, const u8 *src,
 	node->bitlen = bits;
 	memcpy(node->bits, src, bits / 8U);
 }
-#define CHOOSE_NODE(parent, key) \
-	parent->bit[(key[parent->bit_at_a] >> parent->bit_at_b) & 1]
+
+static inline u8 choose(struct allowedips_node *node, const u8 *key)
+{
+	return (key[node->bit_at_a] >> node->bit_at_b) & 1;
+}
 
 static void push_rcu(struct allowedips_node **stack,
 		     struct allowedips_node __rcu *p, unsigned int *len)
@@ -40,6 +45,11 @@ static void push_rcu(struct allowedips_node **stack,
 	}
 }
 
+static void node_free_rcu(struct rcu_head *rcu)
+{
+	kmem_cache_free(node_cache, container_of(rcu, struct allowedips_node, rcu));
+}
+
 static void root_free_rcu(struct rcu_head *rcu)
 {
 	struct allowedips_node *node, *stack[128] = {
@@ -49,7 +59,7 @@ static void root_free_rcu(struct rcu_head *rcu)
 	while (len > 0 && (node = stack[--len])) {
 		push_rcu(stack, node->bit[0], &len);
 		push_rcu(stack, node->bit[1], &len);
-		kfree(node);
+		kmem_cache_free(node_cache, node);
 	}
 }
 
@@ -66,60 +76,6 @@ static void root_remove_peer_lists(struct allowedips_node *root)
 	}
 }
 
-static void walk_remove_by_peer(struct allowedips_node __rcu **top,
-				struct wg_peer *peer, struct mutex *lock)
-{
-#define REF(p) rcu_access_pointer(p)
-#define DEREF(p) rcu_dereference_protected(*(p), lockdep_is_held(lock))
-#define PUSH(p) ({                                                             \
-		WARN_ON(IS_ENABLED(DEBUG) && len >= 128);                      \
-		stack[len++] = p;                                              \
-	})
-
-	struct allowedips_node __rcu **stack[128], **nptr;
-	struct allowedips_node *node, *prev;
-	unsigned int len;
-
-	if (unlikely(!peer || !REF(*top)))
-		return;
-
-	for (prev = NULL, len = 0, PUSH(top); len > 0; prev = node) {
-		nptr = stack[len - 1];
-		node = DEREF(nptr);
-		if (!node) {
-			--len;
-			continue;
-		}
-		if (!prev || REF(prev->bit[0]) == node ||
-		    REF(prev->bit[1]) == node) {
-			if (REF(node->bit[0]))
-				PUSH(&node->bit[0]);
-			else if (REF(node->bit[1]))
-				PUSH(&node->bit[1]);
-		} else if (REF(node->bit[0]) == prev) {
-			if (REF(node->bit[1]))
-				PUSH(&node->bit[1]);
-		} else {
-			if (rcu_dereference_protected(node->peer,
-				lockdep_is_held(lock)) == peer) {
-				RCU_INIT_POINTER(node->peer, NULL);
-				list_del_init(&node->peer_list);
-				if (!node->bit[0] || !node->bit[1]) {
-					rcu_assign_pointer(*nptr, DEREF(
-					       &node->bit[!REF(node->bit[0])]));
-					kfree_rcu(node, rcu);
-					node = DEREF(nptr);
-				}
-			}
-			--len;
-		}
-	}
-
-#undef REF
-#undef DEREF
-#undef PUSH
-}
-
 static unsigned int fls128(u64 a, u64 b)
 {
 	return a ? fls64(a) + 64U : fls64(b);
@@ -159,7 +115,7 @@ static struct allowedips_node *find_node(struct allowedips_node *trie, u8 bits,
 			found = node;
 		if (node->cidr == bits)
 			break;
-		node = rcu_dereference_bh(CHOOSE_NODE(node, key));
+		node = rcu_dereference_bh(node->bit[choose(node, key)]);
 	}
 	return found;
 }
@@ -191,8 +147,7 @@ static bool node_placement(struct allowedips_node __rcu *trie, const u8 *key,
 			   u8 cidr, u8 bits, struct allowedips_node **rnode,
 			   struct mutex *lock)
 {
-	struct allowedips_node *node = rcu_dereference_protected(trie,
-						lockdep_is_held(lock));
+	struct allowedips_node *node = rcu_dereference_protected(trie, lockdep_is_held(lock));
 	struct allowedips_node *parent = NULL;
 	bool exact = false;
 
@@ -202,13 +157,24 @@ static bool node_placement(struct allowedips_node __rcu *trie, const u8 *key,
 			exact = true;
 			break;
 		}
-		node = rcu_dereference_protected(CHOOSE_NODE(parent, key),
-						 lockdep_is_held(lock));
+		node = rcu_dereference_protected(parent->bit[choose(parent, key)], lockdep_is_held(lock));
 	}
 	*rnode = parent;
 	return exact;
 }
 
+static inline void connect_node(struct allowedips_node **parent, u8 bit, struct allowedips_node *node)
+{
+	node->parent_bit_packed = (unsigned long)parent | bit;
+	rcu_assign_pointer(*parent, node);
+}
+
+static inline void choose_and_connect_node(struct allowedips_node *parent, struct allowedips_node *node)
+{
+	u8 bit = choose(parent, node->bits);
+	connect_node(&parent->bit[bit], bit, node);
+}
+
 static int add(struct allowedips_node __rcu **trie, u8 bits, const u8 *key,
 	       u8 cidr, struct wg_peer *peer, struct mutex *lock)
 {
@@ -218,13 +184,13 @@ static int add(struct allowedips_node __rcu **trie, u8 bits, const u8 *key,
 		return -EINVAL;
 
 	if (!rcu_access_pointer(*trie)) {
-		node = kzalloc(sizeof(*node), GFP_KERNEL);
+		node = kmem_cache_zalloc(node_cache, GFP_KERNEL);
 		if (unlikely(!node))
 			return -ENOMEM;
 		RCU_INIT_POINTER(node->peer, peer);
 		list_add_tail(&node->peer_list, &peer->allowedips_list);
 		copy_and_assign_cidr(node, key, cidr, bits);
-		rcu_assign_pointer(*trie, node);
+		connect_node(trie, 2, node);
 		return 0;
 	}
 	if (node_placement(*trie, key, cidr, bits, &node, lock)) {
@@ -233,7 +199,7 @@ static int add(struct allowedips_node __rcu **trie, u8 bits, const u8 *key,
 		return 0;
 	}
 
-	newnode = kzalloc(sizeof(*newnode), GFP_KERNEL);
+	newnode = kmem_cache_zalloc(node_cache, GFP_KERNEL);
 	if (unlikely(!newnode))
 		return -ENOMEM;
 	RCU_INIT_POINTER(newnode->peer, peer);
@@ -243,10 +209,10 @@ static int add(struct allowedips_node __rcu **trie, u8 bits, const u8 *key,
 	if (!node) {
 		down = rcu_dereference_protected(*trie, lockdep_is_held(lock));
 	} else {
-		down = rcu_dereference_protected(CHOOSE_NODE(node, key),
-						 lockdep_is_held(lock));
+		const u8 bit = choose(node, key);
+		down = rcu_dereference_protected(node->bit[bit], lockdep_is_held(lock));
 		if (!down) {
-			rcu_assign_pointer(CHOOSE_NODE(node, key), newnode);
+			connect_node(&node->bit[bit], bit, newnode);
 			return 0;
 		}
 	}
@@ -254,30 +220,29 @@ static int add(struct allowedips_node __rcu **trie, u8 bits, const u8 *key,
 	parent = node;
 
 	if (newnode->cidr == cidr) {
-		rcu_assign_pointer(CHOOSE_NODE(newnode, down->bits), down);
+		choose_and_connect_node(newnode, down);
 		if (!parent)
-			rcu_assign_pointer(*trie, newnode);
+			connect_node(trie, 2, newnode);
 		else
-			rcu_assign_pointer(CHOOSE_NODE(parent, newnode->bits),
-					   newnode);
-	} else {
-		node = kzalloc(sizeof(*node), GFP_KERNEL);
-		if (unlikely(!node)) {
-			list_del(&newnode->peer_list);
-			kfree(newnode);
-			return -ENOMEM;
-		}
-		INIT_LIST_HEAD(&node->peer_list);
-		copy_and_assign_cidr(node, newnode->bits, cidr, bits);
+			choose_and_connect_node(parent, newnode);
+		return 0;
+	}
 
-		rcu_assign_pointer(CHOOSE_NODE(node, down->bits), down);
-		rcu_assign_pointer(CHOOSE_NODE(node, newnode->bits), newnode);
-		if (!parent)
-			rcu_assign_pointer(*trie, node);
-		else
-			rcu_assign_pointer(CHOOSE_NODE(parent, node->bits),
-					   node);
+	node = kmem_cache_zalloc(node_cache, GFP_KERNEL);
+	if (unlikely(!node)) {
+		list_del(&newnode->peer_list);
+		kmem_cache_free(node_cache, newnode);
+		return -ENOMEM;
 	}
+	INIT_LIST_HEAD(&node->peer_list);
+	copy_and_assign_cidr(node, newnode->bits, cidr, bits);
+
+	choose_and_connect_node(node, down);
+	choose_and_connect_node(node, newnode);
+	if (!parent)
+		connect_node(trie, 2, node);
+	else
+		choose_and_connect_node(parent, node);
 	return 0;
 }
 
@@ -335,9 +300,41 @@ int wg_allowedips_insert_v6(struct allowedips *table, const struct in6_addr *ip,
 void wg_allowedips_remove_by_peer(struct allowedips *table,
 				  struct wg_peer *peer, struct mutex *lock)
 {
+	struct allowedips_node *node, *child, **parent_bit, *parent, *tmp;
+	bool free_parent;
+
+	if (list_empty(&peer->allowedips_list))
+		return;
 	++table->seq;
-	walk_remove_by_peer(&table->root4, peer, lock);
-	walk_remove_by_peer(&table->root6, peer, lock);
+	list_for_each_entry_safe(node, tmp, &peer->allowedips_list, peer_list) {
+		list_del_init(&node->peer_list);
+		RCU_INIT_POINTER(node->peer, NULL);
+		if (node->bit[0] && node->bit[1])
+			continue;
+		child = rcu_dereference_protected(node->bit[!rcu_access_pointer(node->bit[0])],
+						  lockdep_is_held(lock));
+		if (child)
+			child->parent_bit_packed = node->parent_bit_packed;
+		parent_bit = (struct allowedips_node **)(node->parent_bit_packed & ~3UL);
+		*parent_bit = child;
+		parent = (void *)parent_bit -
+			 offsetof(struct allowedips_node, bit[node->parent_bit_packed & 1]);
+		free_parent = !rcu_access_pointer(node->bit[0]) &&
+			      !rcu_access_pointer(node->bit[1]) &&
+			      (node->parent_bit_packed & 3) <= 1 &&
+			      !rcu_access_pointer(parent->peer);
+		if (free_parent)
+			child = rcu_dereference_protected(
+					parent->bit[!(node->parent_bit_packed & 1)],
+					lockdep_is_held(lock));
+		call_rcu(&node->rcu, node_free_rcu);
+		if (!free_parent)
+			continue;
+		if (child)
+			child->parent_bit_packed = parent->parent_bit_packed;
+		*(struct allowedips_node **)(parent->parent_bit_packed & ~3UL) = child;
+		call_rcu(&parent->rcu, node_free_rcu);
+	}
 }
 
 int wg_allowedips_read_node(struct allowedips_node *node, u8 ip[16], u8 *cidr)
@@ -374,4 +371,16 @@ struct wg_peer *wg_allowedips_lookup_src(struct allowedips *table,
 	return NULL;
 }
 
+int __init wg_allowedips_slab_init(void)
+{
+	node_cache = KMEM_CACHE(allowedips_node, 0);
+	return node_cache ? 0 : -ENOMEM;
+}
+
+void wg_allowedips_slab_uninit(void)
+{
+	rcu_barrier();
+	kmem_cache_destroy(node_cache);
+}
+
 #include "selftest/allowedips.c"