[PATCH 5.15.y] net/packet: defer vmalloc TX_RING free until skbs finish

[Sasha Levin <sashal@kernel.org>]

From: Kyle Zeng <kylebot@openai.com>

[ Upstream commit 992cc9f94ca924089a506ba9b327caa9af797529 ]

AF_PACKET TX_RING skbs keep a raw pointer to their ring frame. The skb
page references preserve page-backed ring blocks after pg_vec is freed,
but they do not preserve a vmalloc mapping.

tpacket_destruct_skb() currently drops the pending reference before
writing the timestamp and TP_STATUS_AVAILABLE to the frame. Move the
decrement after those stores. The smp_wmb() in __packet_set_status()
orders the frame stores before the decrement.

Also recheck pending TX frames under pg_vec_lock before non-closing
ring replacement, so a racing send cannot add a pending skb between
the initial check and the ring swap.

Ring allocation can produce a mixture of page-backed and vmalloc-backed
blocks. Allocate deferred-work storage during TX ring setup when the
first vmalloc-backed block is encountered, and keep its pointer in the
pg_vec allocation header. If allocation fails, return -ENOMEM from ring
setup. On socket close, a non-NULL pointer identifies a vmalloc-backed
vector without a scan. If TX skbs remain, defer the whole vector to
system_long_wq.

After pg_vec is detached, a late destructor can skip the pending
decrement. Use socket write-memory accounting as the deferred lifetime
gate instead: an skb remains charged through its final sock_wfree(),
after all ring-frame accesses. The delayed work retains a socket
reference and reschedules itself until no TX skbs remain.

Move pending_refcnt release to packet_sock_destruct() so late skb
destructors and deferred cleanup can safely use it after
packet_release(). Page-backed teardown remains synchronous, and no lock
is added to the TX completion hot path.

Fixes: b013840810c2 ("packet: use percpu mmap tx frame pending refcount")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/netdev/20260721015824.45829-1-kylebot@openai.com/
Suggested-by: Eric Dumazet <edumazet@google.com>
Suggested-by: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Kyle Zeng <kylebot@openai.com>
Link: https://patch.msgid.link/20260816235646.76500-1-kylebot@openai.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
[ adapted newer allocation, container, and counted-by helpers to Linux 5.15 equivalents while preserving sk_refcnt_debug_release() ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 net/packet/af_packet.c | 97 ++++++++++++++++++++++++++++++++++++++----
 1 file changed, 88 insertions(+), 9 deletions(-)

diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 447c29d942e9e..232107e611b61 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -86,6 +86,7 @@
 #include <linux/errqueue.h>
 #include <linux/net_tstamp.h>
 #include <linux/percpu.h>
+#include <linux/workqueue.h>
 #ifdef CONFIG_INET
 #include <net/inet_common.h>
 #endif
@@ -1359,6 +1360,8 @@ static void packet_sock_destruct(struct sock *sk)
 	WARN_ON(atomic_read(&sk->sk_rmem_alloc));
 	WARN_ON(refcount_read(&sk->sk_wmem_alloc));
 
+	packet_free_pending(pkt_sk(sk));
+
 	if (!sock_flag(sk, SOCK_DEAD)) {
 		pr_err("Attempt to release alive packet socket: %p\n", sk);
 		return;
@@ -2555,11 +2558,11 @@ static void tpacket_destruct_skb(struct sk_buff *skb)
 		__u32 ts;
 
 		ph = skb_zcopy_get_nouarg(skb);
-		packet_dec_pending(&po->tx_ring);
 
 		ts = __packet_set_timestamp(po, ph, skb);
 		__packet_set_status(po, ph, TP_STATUS_AVAILABLE | ts);
 
+		packet_dec_pending(&po->tx_ring);
 		complete(&po->skb_completion);
 	}
 
@@ -3218,7 +3221,6 @@ static int packet_release(struct socket *sock)
 	/* Purge queues */
 
 	skb_queue_purge(&sk->sk_receive_queue);
-	packet_free_pending(po);
 	sk_refcnt_debug_release(sk);
 
 	sock_put(sk);
@@ -4357,11 +4359,26 @@ static const struct vm_operations_struct packet_mmap_ops = {
 	.close	=	packet_mm_close,
 };
 
+struct packet_pg_vec {
+	struct packet_pg_vec_free *deferred;
+	unsigned int order;
+	unsigned int len;
+	struct pgv pg_vec[];
+};
+
+struct packet_pg_vec_free {
+	struct delayed_work work;
+	struct sock *sk;
+	struct packet_pg_vec *vec;
+};
+
 static void free_pg_vec(struct pgv *pg_vec, unsigned int order,
 			unsigned int len)
 {
+	struct packet_pg_vec *vec;
 	int i;
 
+	vec = container_of(pg_vec, struct packet_pg_vec, pg_vec[0]);
 	for (i = 0; i < len; i++) {
 		if (likely(pg_vec[i].buffer)) {
 			if (is_vmalloc_addr(pg_vec[i].buffer))
@@ -4372,7 +4389,46 @@ static void free_pg_vec(struct pgv *pg_vec, unsigned int order,
 			pg_vec[i].buffer = NULL;
 		}
 	}
-	kfree(pg_vec);
+	kfree(vec->deferred);
+	kfree(vec);
+}
+
+static void packet_free_pg_vec_work(struct work_struct *work)
+{
+	struct packet_pg_vec_free *deferred;
+	struct packet_pg_vec *vec;
+	struct sock *sk;
+
+	deferred = container_of(to_delayed_work(work),
+				struct packet_pg_vec_free, work);
+	vec = deferred->vec;
+	sk = deferred->sk;
+	if (sk_wmem_alloc_get(sk)) {
+		queue_delayed_work(system_long_wq, &deferred->work, 1);
+		return;
+	}
+
+	free_pg_vec(vec->pg_vec, vec->order, vec->len);
+	sock_put(sk);
+}
+
+static void packet_free_tx_ring(struct sock *sk, struct pgv *pg_vec,
+				unsigned int order, unsigned int len)
+{
+	struct packet_pg_vec_free *deferred;
+	struct packet_pg_vec *vec;
+
+	vec = container_of(pg_vec, struct packet_pg_vec, pg_vec[0]);
+	deferred = vec->deferred;
+	if (!deferred || !sk_wmem_alloc_get(sk)) {
+		free_pg_vec(pg_vec, order, len);
+		return;
+	}
+
+	/* A detached ring's pending count can miss late skb destructors. */
+	deferred->sk = sk;
+	sock_hold(sk);
+	queue_delayed_work(system_long_wq, &deferred->work, 0);
 }
 
 static char *alloc_one_pg_vec_page(unsigned long order)
@@ -4400,20 +4456,36 @@ static char *alloc_one_pg_vec_page(unsigned long order)
 	return NULL;
 }
 
-static struct pgv *alloc_pg_vec(struct tpacket_req *req, int order)
+static struct pgv *alloc_pg_vec(struct tpacket_req *req, int order, bool tx_ring)
 {
 	unsigned int block_nr = req->tp_block_nr;
+	struct packet_pg_vec *vec;
 	struct pgv *pg_vec;
 	int i;
 
-	pg_vec = kcalloc(block_nr, sizeof(struct pgv), GFP_KERNEL | __GFP_NOWARN);
-	if (unlikely(!pg_vec))
-		goto out;
+	vec = kzalloc(struct_size(vec, pg_vec, block_nr),
+		      GFP_KERNEL | __GFP_NOWARN);
+	if (unlikely(!vec))
+		return NULL;
+	vec->order = order;
+	vec->len = block_nr;
+	pg_vec = vec->pg_vec;
 
 	for (i = 0; i < block_nr; i++) {
 		pg_vec[i].buffer = alloc_one_pg_vec_page(order);
 		if (unlikely(!pg_vec[i].buffer))
 			goto out_free_pgvec;
+
+		if (tx_ring && !vec->deferred &&
+		    is_vmalloc_addr(pg_vec[i].buffer)) {
+			vec->deferred = kzalloc(sizeof(*vec->deferred),
+						GFP_KERNEL | __GFP_NOWARN);
+			if (!vec->deferred)
+				goto out_free_pgvec;
+			vec->deferred->vec = vec;
+			INIT_DELAYED_WORK(&vec->deferred->work,
+					  packet_free_pg_vec_work);
+		}
 	}
 
 out:
@@ -4496,7 +4568,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 
 		err = -ENOMEM;
 		order = get_order(req->tp_block_size);
-		pg_vec = alloc_pg_vec(req, order);
+		pg_vec = alloc_pg_vec(req, order, tx_ring);
 		if (unlikely(!pg_vec))
 			goto out;
 		switch (po->tp_version) {
@@ -4548,6 +4620,9 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 	err = -EBUSY;
 	mutex_lock(&po->pg_vec_lock);
 	if (closing || atomic_long_read(&po->mapped) == 0) {
+		if (tx_ring && !closing && packet_read_pending(rb))
+			goto out_unlock;
+
 		err = 0;
 		spin_lock_bh(&rb_queue->lock);
 		swap(rb->pg_vec, pg_vec);
@@ -4569,6 +4644,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 			pr_err("packet_mmap: vma is busy: %ld\n",
 			       atomic_long_read(&po->mapped));
 	}
+out_unlock:
 	mutex_unlock(&po->pg_vec_lock);
 
 	spin_lock(&po->bind_lock);
@@ -4590,7 +4666,10 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 out_free_pg_vec:
 	if (pg_vec) {
 		bitmap_free(rx_owner_map);
-		free_pg_vec(pg_vec, order, req->tp_block_nr);
+		if (tx_ring && closing)
+			packet_free_tx_ring(sk, pg_vec, order, req->tp_block_nr);
+		else
+			free_pg_vec(pg_vec, order, req->tp_block_nr);
 	}
 out:
 	return err;
-- 
2.53.0