Skip to contentSkip to conversation

FAILED: patch "[PATCH] mptcp: pm: fix memory leak from alloc-during-teardown race" failed to apply to 7.1-stable tree

PATCH New [email protected] 2 participants 4 messages Aug 20 – Aug 24

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 7.1-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-7.1.y
git checkout FETCH_HEAD
git cherry-pick -x efc33b5102ff859bacd390a5f30112d8e0c084c0
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2026082045-blaming-kiln-83fe@gregkh' --subject-prefix 'PATCH 7.1.y' 'HEAD^..'

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From efc33b5102ff859bacd390a5f30112d8e0c084c0 Mon Sep 17 00:00:00 2001
From: Shardul Bankar <[email protected]>
Date: Mon, 3 Aug 2026 18:16:38 +0200
Subject: [PATCH] mptcp: pm: fix memory leak from alloc-during-teardown race

mptcp_pm_destroy() empties msk->pm.anno_list and msk->pm.userspace_pm_local_addr_list under msk->pm.lock during socket teardown, dropping the lock between the two.

A concurrent userspace PM genl ANNOUNCE on the same msk holds a sock reference via mptcp_token_get_sock() and, in mptcp_pm_nl_announce_doit(), calls mptcp_userspace_pm_append_new_local_addr() and mptcp_pm_announced_alloc(). Both take msk->pm.lock briefly to add to their respective lists. Because the genl handler holds a sock reference, mptcp_pm_destroy() may run on the same msk via mptcp_disconnect(), which invokes mptcp_destroy_common() without dropping the sock refcount, before the handler completes.

If the lock acquisitions interleave such that mptcp_pm_destroy() empties a list first, the later alloc adds its entry to a list head that nothing else iterates for this msk, and the entry leaks. kmemleak reports both mptcp_pm_add_addr objects (from mptcp_pm_announced_alloc()) and mptcp_pm_addr_entry objects (from mptcp_userspace_pm_append_new_local_addr()) under sustained concurrent ANNOUNCE + close load against the userspace PM.

Add an MPTCP_PM_DESTROYING bit in msk->pm.status, set by mptcp_pm_destroy() under pm.lock before the lists are emptied and checked under pm.lock by the alloc paths. Either the alloc takes pm.lock first, in which case its entry is on the list when mptcp_pm_destroy() frees it; or mptcp_pm_destroy() takes pm.lock first, in which case the later alloc observes the bit and refuses.

Found by an MPTCP protocol-flow harness extending BRF (arXiv:2305.08782).

net/mptcp/pm.c +11 -2
index c71dcf887683..64a1236aabee 100644
@@ -443,6 +443,9 @@ bool mptcp_pm_announced_alloc(struct mptcp_sock *msk,
 
 	lockdep_assert_held(&msk->pm.lock);
 
+	if (msk->pm.status & BIT(MPTCP_PM_DESTROYING))
+		return false;
+
 	add_entry = mptcp_pm_announced_lookup(msk, addr);
 	if (add_entry) {
 		if (WARN_ON_ONCE(mptcp_pm_is_kernel(msk)))
@@ -1145,10 +1148,16 @@ void mptcp_pm_worker(struct mptcp_sock *msk)
 
 void mptcp_pm_destroy(struct mptcp_sock *msk)
 {
+	spin_lock_bh(&msk->pm.lock);
+	msk->pm.status |= BIT(MPTCP_PM_DESTROYING);
+	spin_unlock_bh(&msk->pm.lock);
+
 	mptcp_pm_free_announced_list(msk);
 
-	if (mptcp_pm_is_userspace(msk))
-		mptcp_userspace_pm_free_local_addr_list(msk);
+	/* Free the userspace local address list unconditionally: the socket
+	 * can be reused (mptcp_disconnect()) and re-selected to a different PM
+	 */
+	mptcp_userspace_pm_free_local_addr_list(msk);
 }
 
 void mptcp_pm_data_reset(struct mptcp_sock *msk)
net/mptcp/pm_userspace.c +4 -0
index 945aa5afc2dd..2203cc2d2748 100644
@@ -54,6 +54,10 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 	bitmap_zero(id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
 
 	spin_lock_bh(&msk->pm.lock);
+	if (msk->pm.status & BIT(MPTCP_PM_DESTROYING)) {
+		ret = -EINVAL;
+		goto append_err;
+	}
 	mptcp_for_each_userspace_pm_addr(msk, e) {
 		addr_match = mptcp_addresses_equal(&e->addr, &entry->addr, true);
 		if (addr_match && entry->addr.id == 0 && needs_id)
net/mptcp/protocol.h +4 -3
index c13680d18994..1b80f2d6ec5a 100644
@@ -190,9 +190,10 @@ enum mptcp_pm_status {
 	MPTCP_PM_ESTABLISHED,
 	MPTCP_PM_SUBFLOW_ESTABLISHED,
 	MPTCP_PM_ALREADY_ESTABLISHED,	/* persistent status, set after ESTABLISHED event */
-	MPTCP_PM_MPC_ENDPOINT_ACCOUNTED /* persistent status, set after MPC local address is
-					 * accounted int id_avail_bitmap
-					 */
+	MPTCP_PM_MPC_ENDPOINT_ACCOUNTED, /* persistent status, set after MPC local address is
+					  * accounted int id_avail_bitmap
+					  */
+	MPTCP_PM_DESTROYING,		/* To fence out PM list allocs */
 };
 
 enum mptcp_pm_type {
maintainer
#1 lore raw

PATCH 1/3 mptcp: pm: rename add_entry structure to add_addr

From: "Matthieu Baerts (NGI0)" <[email protected]>

[ Upstream commit 350d76dd6e79468ac85767f2d236299a135572df ]

Using only the 'add' prefix is confusing: does it refer to a generic added entry or address, or specifically to ADD_ADDRs. Using add_addr removes this confusion.

Stable-dep-of: efc33b5102ff ("mptcp: pm: fix memory leak from alloc-during-teardown race")

---
 net/mptcp/pm.c       | 22 +++++++++++-----------
 net/mptcp/protocol.h |  2 +-
 2 files changed, 12 insertions(+), 12 deletions(-)
net/mptcp/pm.c +11 -11
index c5f95c251fbfa..745ccca474db7 100644
@@ -12,7 +12,7 @@
 
 #define ADD_ADDR_RETRANS_MAX	3
 
-struct mptcp_pm_add_entry {
+struct mptcp_pm_add_addr {
 	struct list_head	list;
 	struct mptcp_addr_info	addr;
 	u8			retrans_times;
@@ -133,11 +133,11 @@ bool mptcp_lookup_subflow_by_saddr(const struct list_head *list,
 	return false;
 }
 
-static struct mptcp_pm_add_entry *
+static struct mptcp_pm_add_addr *
 mptcp_lookup_anno_list_by_saddr(const struct mptcp_sock *msk,
 				const struct mptcp_addr_info *addr)
 {
-	struct mptcp_pm_add_entry *entry;
+	struct mptcp_pm_add_addr *entry;
 
 	lockdep_assert_held(&msk->pm.lock);
 
@@ -152,7 +152,7 @@ mptcp_lookup_anno_list_by_saddr(const struct mptcp_sock *msk,
 bool mptcp_remove_anno_list_by_saddr(struct mptcp_sock *msk,
 				     const struct mptcp_addr_info *addr)
 {
-	struct mptcp_pm_add_entry *entry;
+	struct mptcp_pm_add_addr *entry;
 	bool ret;
 
 	entry = mptcp_pm_del_add_timer(msk, addr, false);
@@ -164,7 +164,7 @@ bool mptcp_remove_anno_list_by_saddr(struct mptcp_sock *msk,
 
 bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk)
 {
-	struct mptcp_pm_add_entry *entry;
+	struct mptcp_pm_add_addr *entry;
 	struct mptcp_addr_info saddr;
 	bool ret = false;
 
@@ -340,8 +340,8 @@ static unsigned int mptcp_adjust_add_addr_timeout(struct mptcp_sock *msk)
 
 static void mptcp_pm_add_timer(struct timer_list *timer)
 {
-	struct mptcp_pm_add_entry *entry = timer_container_of(entry, timer,
-							      add_timer);
+	struct mptcp_pm_add_addr *entry = timer_container_of(entry, timer,
+							     add_timer);
 	struct mptcp_sock *msk = entry->sock;
 	struct sock *sk = (struct sock *)msk;
 	unsigned int timeout = 0;
@@ -399,12 +399,12 @@ static void mptcp_pm_add_timer(struct timer_list *timer)
 	sock_put(sk);
 }
 
-struct mptcp_pm_add_entry *
+struct mptcp_pm_add_addr *
 mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 		       const struct mptcp_addr_info *addr, bool check_id)
 {
-	struct mptcp_pm_add_entry *entry;
 	struct sock *sk = (struct sock *)msk;
+	struct mptcp_pm_add_addr *entry;
 	bool stop_timer = false;
 
 	rcu_read_lock();
@@ -436,7 +436,7 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 			      const struct mptcp_addr_info *addr)
 {
-	struct mptcp_pm_add_entry *add_entry = NULL;
+	struct mptcp_pm_add_addr *add_entry = NULL;
 	struct sock *sk = (struct sock *)msk;
 	unsigned int timeout;
 
@@ -473,7 +473,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 
 static void mptcp_pm_free_anno_list(struct mptcp_sock *msk)
 {
-	struct mptcp_pm_add_entry *entry, *tmp;
+	struct mptcp_pm_add_addr *entry, *tmp;
 	struct sock *sk = (struct sock *)msk;
 	LIST_HEAD(free_list);
 
net/mptcp/protocol.h +1 -1
index 352d5771de0b1..3c5dd07f57bb0 100644
@@ -1133,7 +1133,7 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk,
 bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 			      const struct mptcp_addr_info *addr);
 bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk);
-struct mptcp_pm_add_entry *
+struct mptcp_pm_add_addr *
 mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 		       const struct mptcp_addr_info *addr, bool check_id);
 bool mptcp_lookup_subflow_by_saddr(const struct list_head *list,
Signature · 1 line
2.53.0
maintainer
#2 lore raw

PATCH 2/3 mptcp: pm: uniform announced addresses helpers

From: "Matthieu Baerts (NGI0)" <[email protected]>

[ Upstream commit 7d4dacc8ccca8e40fcb5f2ece9ea3e502da1bd3f ]

Similar to the previous commit, only using the 'add' or 'anno' prefixes is confusing -- generally associated to the action of adding something, or the Latin name for "year" -- and lack of uniformity.

This has been causing issues in the past, e.g. del_add_timer seemed to suggest the goal is to delete a previously added timer.

Instead, use the mptcp_pm_announced_ prefix.

While at it, slightly improves some helpers:

- mptcp_lookup_anno_list_by_saddr: no need to specify what is used to do the lookup: mptcp_pm_announced_lookup.

- mptcp_pm_sport_in_anno_list: it doesn't just compare the port, but the whole address linked to the sublow: mptcp_pm_announced_has_ssk.

- mptcp_pm_alloc_anno_list: it allocates one item of the list, not a whole list: mptcp_pm_announced_alloc.

Stable-dep-of: efc33b5102ff ("mptcp: pm: fix memory leak from alloc-during-teardown race")

---
 net/mptcp/options.c      |  2 +-
 net/mptcp/pm.c           | 35 +++++++++++++++++------------------
 net/mptcp/pm_kernel.c    | 10 +++++-----
 net/mptcp/pm_userspace.c |  4 ++--
 net/mptcp/protocol.h     | 12 ++++++------
 net/mptcp/subflow.c      |  4 ++--
 6 files changed, 33 insertions(+), 34 deletions(-)
net/mptcp/options.c +1 -1
index 6c8b214e76591..36517ab14d410 100644
@@ -1261,7 +1261,7 @@ bool mptcp_incoming_options(struct sock *sk, struct sk_buff *skb)
 				MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_ADDADDR);
 			} else {
 				mptcp_pm_add_addr_echoed(msk, &mp_opt.addr);
-				mptcp_pm_del_add_timer(msk, &mp_opt.addr, true);
+				mptcp_pm_announced_del_timer(msk, &mp_opt.addr, true);
 				MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_ECHOADD);
 			}
 
net/mptcp/pm.c +17 -18
index 745ccca474db7..3d1c4d6f517d8 100644
@@ -134,8 +134,8 @@ bool mptcp_lookup_subflow_by_saddr(const struct list_head *list,
 }
 
 static struct mptcp_pm_add_addr *
-mptcp_lookup_anno_list_by_saddr(const struct mptcp_sock *msk,
-				const struct mptcp_addr_info *addr)
+mptcp_pm_announced_lookup(const struct mptcp_sock *msk,
+			  const struct mptcp_addr_info *addr)
 {
 	struct mptcp_pm_add_addr *entry;
 
@@ -149,26 +149,26 @@ mptcp_lookup_anno_list_by_saddr(const struct mptcp_sock *msk,
 	return NULL;
 }
 
-bool mptcp_remove_anno_list_by_saddr(struct mptcp_sock *msk,
-				     const struct mptcp_addr_info *addr)
+bool mptcp_pm_announced_remove(struct mptcp_sock *msk,
+			       const struct mptcp_addr_info *addr)
 {
 	struct mptcp_pm_add_addr *entry;
 	bool ret;
 
-	entry = mptcp_pm_del_add_timer(msk, addr, false);
+	entry = mptcp_pm_announced_del_timer(msk, addr, false);
 	ret = entry;
 	kfree_rcu(entry, rcu);
 
 	return ret;
 }
 
-bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk)
+bool mptcp_pm_announced_has_ssk(struct mptcp_sock *msk, const struct sock *ssk)
 {
 	struct mptcp_pm_add_addr *entry;
 	struct mptcp_addr_info saddr;
 	bool ret = false;
 
-	mptcp_local_address((struct sock_common *)sk, &saddr);
+	mptcp_local_address((struct sock_common *)ssk, &saddr);
 
 	spin_lock_bh(&msk->pm.lock);
 	list_for_each_entry(entry, &msk->pm.anno_list, list) {
@@ -365,7 +365,7 @@ static void mptcp_pm_add_timer(struct timer_list *timer)
 
 	spin_lock_bh(&msk->pm.lock);
 
-	/* The cancel path (mptcp_pm_del_add_timer()) can race with this
+	/* The cancel path (mptcp_pm_announced_del_timer()) can race with this
 	 * callback. Once cancel updates retrans_times to MAX, suppress further
 	 * retransmissions here. If this callback acquires pm.lock first, one
 	 * final transmit attempt is still possible.
@@ -400,8 +400,8 @@ static void mptcp_pm_add_timer(struct timer_list *timer)
 }
 
 struct mptcp_pm_add_addr *
-mptcp_pm_del_add_timer(struct mptcp_sock *msk,
-		       const struct mptcp_addr_info *addr, bool check_id)
+mptcp_pm_announced_del_timer(struct mptcp_sock *msk,
+			     const struct mptcp_addr_info *addr, bool check_id)
 {
 	struct sock *sk = (struct sock *)msk;
 	struct mptcp_pm_add_addr *entry;
@@ -410,7 +410,7 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 	rcu_read_lock();
 
 	spin_lock_bh(&msk->pm.lock);
-	entry = mptcp_lookup_anno_list_by_saddr(msk, addr);
+	entry = mptcp_pm_announced_lookup(msk, addr);
 	if (entry && (!check_id || entry->addr.id == addr->id)) {
 		entry->retrans_times = ADD_ADDR_RETRANS_MAX;
 		stop_timer = true;
@@ -433,7 +433,7 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk,
 	return entry;
 }
 
-bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
+bool mptcp_pm_announced_alloc(struct mptcp_sock *msk,
 			      const struct mptcp_addr_info *addr)
 {
 	struct mptcp_pm_add_addr *add_entry = NULL;
@@ -442,8 +442,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 
 	lockdep_assert_held(&msk->pm.lock);
 
-	add_entry = mptcp_lookup_anno_list_by_saddr(msk, addr);
-
+	add_entry = mptcp_pm_announced_lookup(msk, addr);
 	if (add_entry) {
 		if (WARN_ON_ONCE(mptcp_pm_is_kernel(msk)))
 			return false;
@@ -471,7 +470,7 @@ bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
 	return true;
 }
 
-static void mptcp_pm_free_anno_list(struct mptcp_sock *msk)
+static void mptcp_pm_free_announced_list(struct mptcp_sock *msk)
 {
 	struct mptcp_pm_add_addr *entry, *tmp;
 	struct sock *sk = (struct sock *)msk;
@@ -740,7 +739,7 @@ void mptcp_pm_add_addr_echoed(struct mptcp_sock *msk,
 
 	spin_lock_bh(&pm->lock);
 
-	if (mptcp_lookup_anno_list_by_saddr(msk, addr) && READ_ONCE(pm->work_pending))
+	if (mptcp_pm_announced_lookup(msk, addr) && READ_ONCE(pm->work_pending))
 		mptcp_pm_schedule_work(msk, MPTCP_PM_SUBFLOW_ESTABLISHED);
 
 	spin_unlock_bh(&pm->lock);
@@ -949,7 +948,7 @@ bool mptcp_pm_add_addr_signal(struct mptcp_sock *msk, unsigned int opt_size,
 	 * let the PM state machine progress.
 	 */
 	if (skip_add_addr) {
-		mptcp_pm_del_add_timer(msk, addr, true);
+		mptcp_pm_announced_del_timer(msk, addr, true);
 		mptcp_pm_subflow_established(msk);
 	}
 	return ret;
@@ -1104,7 +1103,7 @@ void mptcp_pm_worker(struct mptcp_sock *msk)
 
 void mptcp_pm_destroy(struct mptcp_sock *msk)
 {
-	mptcp_pm_free_anno_list(msk);
+	mptcp_pm_free_announced_list(msk);
 
 	if (mptcp_pm_is_userspace(msk))
 		mptcp_userspace_pm_free_local_addr_list(msk);
net/mptcp/pm_kernel.c +5 -5
index fc818b63752e3..0c5f8d3e35a5a 100644
@@ -370,7 +370,7 @@ static void mptcp_pm_create_subflow_or_signal_addr(struct mptcp_sock *msk)
 		/* If the alloc fails, we are on memory pressure, not worth
 		 * continuing, and trying to create subflows.
 		 */
-		if (!mptcp_pm_alloc_anno_list(msk, &local.addr))
+		if (!mptcp_pm_announced_alloc(msk, &local.addr))
 			return;
 
 		__clear_bit(endp_id, msk->pm.id_avail_bitmap);
@@ -1042,7 +1042,7 @@ int mptcp_pm_nl_add_addr_doit(struct sk_buff *skb, struct genl_info *info)
 	return ret;
 }
 
-static void mptcp_pm_remove_anno_addr(struct mptcp_sock *msk,
+static void mptcp_pm_remove_announced(struct mptcp_sock *msk,
 				      const struct mptcp_addr_info *addr,
 				      bool force)
 {
@@ -1051,7 +1051,7 @@ static void mptcp_pm_remove_anno_addr(struct mptcp_sock *msk,
 
 	list.ids[list.nr++] = mptcp_endp_get_local_id(msk, addr);
 
-	announced = mptcp_remove_anno_list_by_saddr(msk, addr);
+	announced = mptcp_pm_announced_remove(msk, addr);
 	if (announced || force) {
 		spin_lock_bh(&msk->pm.lock);
 		if (announced)
@@ -1088,7 +1088,7 @@ static int mptcp_nl_remove_subflow_and_signal_addr(struct net *net,
 
 		lock_sock(sk);
 		remove_subflow = mptcp_lookup_subflow_by_saddr(&msk->conn_list, addr);
-		mptcp_pm_remove_anno_addr(msk, addr, remove_subflow &&
+		mptcp_pm_remove_announced(msk, addr, remove_subflow &&
 					  !(entry->flags & MPTCP_PM_ADDR_FLAG_IMPLICIT));
 
 		list.ids[0] = mptcp_endp_get_local_id(msk, addr);
@@ -1223,7 +1223,7 @@ static void mptcp_pm_flush_addrs_and_subflows(struct mptcp_sock *msk,
 			slist.ids[slist.nr++] = mptcp_endp_get_local_id(msk, &entry->addr);
 
 		if (alist.nr < MPTCP_RM_IDS_MAX &&
-		    mptcp_remove_anno_list_by_saddr(msk, &entry->addr))
+		    mptcp_pm_announced_remove(msk, &entry->addr))
 			alist.ids[alist.nr++] = mptcp_endp_get_local_id(msk, &entry->addr);
 	}
 
net/mptcp/pm_userspace.c +2 -2
index db5693262c8c6..f2ca04c58978d 100644
@@ -231,7 +231,7 @@ int mptcp_pm_nl_announce_doit(struct sk_buff *skb, struct genl_info *info)
 	lock_sock(sk);
 	spin_lock_bh(&msk->pm.lock);
 
-	if (mptcp_pm_alloc_anno_list(msk, &addr_val.addr)) {
+	if (mptcp_pm_announced_alloc(msk, &addr_val.addr)) {
 		msk->pm.add_addr_signaled++;
 		mptcp_pm_announce_addr(msk, &addr_val.addr, false);
 		mptcp_pm_addr_send_ack(msk);
@@ -284,7 +284,7 @@ void mptcp_pm_remove_addr_entry(struct mptcp_sock *msk,
 	int anno_nr = 0;
 
 	/* only delete if either announced or matching a subflow */
-	if (mptcp_remove_anno_list_by_saddr(msk, &entry->addr))
+	if (mptcp_pm_announced_remove(msk, &entry->addr))
 		anno_nr++;
 	else if (!mptcp_lookup_subflow_by_saddr(&msk->conn_list, &entry->addr))
 		return;
net/mptcp/protocol.h +6 -6
index 3c5dd07f57bb0..b0e6af2cd8ac6 100644
@@ -1130,16 +1130,16 @@ int mptcp_pm_mp_prio_send_ack(struct mptcp_sock *msk,
 			      struct mptcp_addr_info *addr,
 			      struct mptcp_addr_info *rem,
 			      u8 bkup);
-bool mptcp_pm_alloc_anno_list(struct mptcp_sock *msk,
+bool mptcp_pm_announced_alloc(struct mptcp_sock *msk,
 			      const struct mptcp_addr_info *addr);
-bool mptcp_pm_sport_in_anno_list(struct mptcp_sock *msk, const struct sock *sk);
 struct mptcp_pm_add_addr *
-mptcp_pm_del_add_timer(struct mptcp_sock *msk,
-		       const struct mptcp_addr_info *addr, bool check_id);
+mptcp_pm_announced_del_timer(struct mptcp_sock *msk,
+			     const struct mptcp_addr_info *addr, bool check_id);
+bool mptcp_pm_announced_remove(struct mptcp_sock *msk,
+			       const struct mptcp_addr_info *addr);
+bool mptcp_pm_announced_has_ssk(struct mptcp_sock *msk, const struct sock *ssk);
 bool mptcp_lookup_subflow_by_saddr(const struct list_head *list,
 				   const struct mptcp_addr_info *saddr);
-bool mptcp_remove_anno_list_by_saddr(struct mptcp_sock *msk,
-				     const struct mptcp_addr_info *addr);
 int mptcp_pm_nl_set_flags(struct mptcp_pm_addr_entry *local,
 			  struct genl_info *info);
 int mptcp_userspace_pm_set_flags(struct mptcp_pm_addr_entry *local,
net/mptcp/subflow.c +2 -2
index d562e149606f6..8e386899ceb98 100644
@@ -235,7 +235,7 @@ static int subflow_check_req(struct request_sock *req,
 			pr_debug("syn inet_sport=%d %d\n",
 				 ntohs(inet_sk(sk_listener)->inet_sport),
 				 ntohs(inet_sk((struct sock *)subflow_req->msk)->inet_sport));
-			if (!mptcp_pm_sport_in_anno_list(subflow_req->msk, sk_listener)) {
+			if (!mptcp_pm_announced_has_ssk(subflow_req->msk, sk_listener)) {
 				SUBFLOW_REQ_INC_STATS(req, MPTCP_MIB_MISMATCHPORTSYNRX);
 				subflow_add_reset_reason(skb, MPTCP_RST_EPROHIBIT);
 				return -EPERM;
@@ -926,7 +926,7 @@ static struct sock *subflow_syn_recv_sock(const struct sock *sk,
 				pr_debug("ack inet_sport=%d %d\n",
 					 ntohs(inet_sk(sk)->inet_sport),
 					 ntohs(inet_sk((struct sock *)owner)->inet_sport));
-				if (!mptcp_pm_sport_in_anno_list(owner, sk)) {
+				if (!mptcp_pm_announced_has_ssk(owner, sk)) {
 					SUBFLOW_REQ_INC_STATS(req, MPTCP_MIB_MISMATCHPORTACKRX);
 					subflow_add_reset_reason(skb, MPTCP_RST_EPROHIBIT);
 					goto dispose_child;
Signature · 1 line
2.53.0
maintainer
#3 lore raw

PATCH 3/3 mptcp: pm: fix memory leak from alloc-during-teardown race

In reply to Sasha Levin

From: Shardul Bankar <[email protected]>

[ Upstream commit efc33b5102ff859bacd390a5f30112d8e0c084c0 ]

mptcp_pm_destroy() empties msk->pm.anno_list and msk->pm.userspace_pm_local_addr_list under msk->pm.lock during socket teardown, dropping the lock between the two.

A concurrent userspace PM genl ANNOUNCE on the same msk holds a sock reference via mptcp_token_get_sock() and, in mptcp_pm_nl_announce_doit(), calls mptcp_userspace_pm_append_new_local_addr() and mptcp_pm_announced_alloc(). Both take msk->pm.lock briefly to add to their respective lists. Because the genl handler holds a sock reference, mptcp_pm_destroy() may run on the same msk via mptcp_disconnect(), which invokes mptcp_destroy_common() without dropping the sock refcount, before the handler completes.

If the lock acquisitions interleave such that mptcp_pm_destroy() empties a list first, the later alloc adds its entry to a list head that nothing else iterates for this msk, and the entry leaks. kmemleak reports both mptcp_pm_add_addr objects (from mptcp_pm_announced_alloc()) and mptcp_pm_addr_entry objects (from mptcp_userspace_pm_append_new_local_addr()) under sustained concurrent ANNOUNCE + close load against the userspace PM.

Add an MPTCP_PM_DESTROYING bit in msk->pm.status, set by mptcp_pm_destroy() under pm.lock before the lists are emptied and checked under pm.lock by the alloc paths. Either the alloc takes pm.lock first, in which case its entry is on the list when mptcp_pm_destroy() frees it; or mptcp_pm_destroy() takes pm.lock first, in which case the later alloc observes the bit and refuses.

Found by an MPTCP protocol-flow harness extending BRF (arXiv:2305.08782).

---
 net/mptcp/pm.c           | 13 +++++++++++--
 net/mptcp/pm_userspace.c |  4 ++++
 net/mptcp/protocol.h     |  7 ++++---
 3 files changed, 19 insertions(+), 5 deletions(-)
net/mptcp/pm.c +11 -2
index 3d1c4d6f517d8..6273dd075d361 100644
@@ -442,6 +442,9 @@ bool mptcp_pm_announced_alloc(struct mptcp_sock *msk,
 
 	lockdep_assert_held(&msk->pm.lock);
 
+	if (msk->pm.status & BIT(MPTCP_PM_DESTROYING))
+		return false;
+
 	add_entry = mptcp_pm_announced_lookup(msk, addr);
 	if (add_entry) {
 		if (WARN_ON_ONCE(mptcp_pm_is_kernel(msk)))
@@ -1103,10 +1106,16 @@ void mptcp_pm_worker(struct mptcp_sock *msk)
 
 void mptcp_pm_destroy(struct mptcp_sock *msk)
 {
+	spin_lock_bh(&msk->pm.lock);
+	msk->pm.status |= BIT(MPTCP_PM_DESTROYING);
+	spin_unlock_bh(&msk->pm.lock);
+
 	mptcp_pm_free_announced_list(msk);
 
-	if (mptcp_pm_is_userspace(msk))
-		mptcp_userspace_pm_free_local_addr_list(msk);
+	/* Free the userspace local address list unconditionally: the socket
+	 * can be reused (mptcp_disconnect()) and re-selected to a different PM
+	 */
+	mptcp_userspace_pm_free_local_addr_list(msk);
 }
 
 void mptcp_pm_data_reset(struct mptcp_sock *msk)
net/mptcp/pm_userspace.c +4 -0
index f2ca04c58978d..24d3a91fd68a7 100644
@@ -54,6 +54,10 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
 	bitmap_zero(id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
 
 	spin_lock_bh(&msk->pm.lock);
+	if (msk->pm.status & BIT(MPTCP_PM_DESTROYING)) {
+		ret = -EINVAL;
+		goto append_err;
+	}
 	mptcp_for_each_userspace_pm_addr(msk, e) {
 		addr_match = mptcp_addresses_equal(&e->addr, &entry->addr, true);
 		if (addr_match && entry->addr.id == 0 && needs_id)
net/mptcp/protocol.h +4 -3
index b0e6af2cd8ac6..c89be36cd3e8e 100644
@@ -190,9 +190,10 @@ enum mptcp_pm_status {
 	MPTCP_PM_ESTABLISHED,
 	MPTCP_PM_SUBFLOW_ESTABLISHED,
 	MPTCP_PM_ALREADY_ESTABLISHED,	/* persistent status, set after ESTABLISHED event */
-	MPTCP_PM_MPC_ENDPOINT_ACCOUNTED /* persistent status, set after MPC local address is
-					 * accounted int id_avail_bitmap
-					 */
+	MPTCP_PM_MPC_ENDPOINT_ACCOUNTED, /* persistent status, set after MPC local address is
+					  * accounted int id_avail_bitmap
+					  */
+	MPTCP_PM_DESTROYING,		/* To fence out PM list allocs */
 };
 
 enum mptcp_pm_type {
Signature · 1 line
2.53.0