sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

alb_break_llc() decides whether to break LLC preference during active
load balance. It does so by testing that every runnable fair task on the
source rq prefers its LLC:

	env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable

But the two counters cover different sets. nr_pref_llc_running is updated
in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued,
so it follows queued tasks. h_nr_runnable is updated in set_delayed()/
clear_delayed() and drops delay-dequeued tasks.

So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted
in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks,
alb_break_llc() returns false, and active balance is free to pull a task
off its preferred LLC. Active balance only moves runnable tasks, and this
is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB
skips the per-task test in can_migrate_task(). The runnable set is the one
we want.

Fix it on the counter side. A task should be counted in
nr_pref_llc_running exactly while it is both queued on its preferred LLC
(pref_llc_queued) and runnable (!sched_delayed). Define that membership
once in task_pref_llc_runnable(), and adjust the counter only through
pref_llc_running_inc()/pref_llc_running_dec() from the four sites that
change either input: account_llc_enqueue(), account_llc_dequeue(),
set_delayed() and clear_delayed(). Gating every update on the same
predicate keeps the delay, wake and dequeue paths from double-counting
or underflowing; see the comments at those sites for the ordering.

nr_llc_running and sd->llc_counts are not touched and stay on queued
semantics.

Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC")
Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/
Reported-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 204 insertions, 0 deletions

diff --git a/drivers/net/wireguard/queueing.h b/drivers/net/wireguard/queueing.h
new file mode 100644
index 000000000..79b6d70de
--- /dev/null
+++ b/drivers/net/wireguard/queueing.h
@@ -0,0 +1,204 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2015-2019 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
+ */
+
+#ifndef _WG_QUEUEING_H
+#define _WG_QUEUEING_H
+
+#include "peer.h"
+#include <linux/types.h>
+#include <linux/skbuff.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <net/ip_tunnels.h>
+
+struct wg_device;
+struct wg_peer;
+struct multicore_worker;
+struct crypt_queue;
+struct prev_queue;
+struct sk_buff;
+
+/* queueing.c APIs: */
+int wg_packet_queue_init(struct crypt_queue *queue, work_func_t function,
+			 unsigned int len);
+void wg_packet_queue_free(struct crypt_queue *queue, bool purge);
+struct multicore_worker __percpu *
+wg_packet_percpu_multicore_worker_alloc(work_func_t function, void *ptr);
+
+/* receive.c APIs: */
+void wg_packet_receive(struct wg_device *wg, struct sk_buff *skb);
+void wg_packet_handshake_receive_worker(struct work_struct *work);
+/* NAPI poll function: */
+int wg_packet_rx_poll(struct napi_struct *napi, int budget);
+/* Workqueue worker: */
+void wg_packet_decrypt_worker(struct work_struct *work);
+
+/* send.c APIs: */
+void wg_packet_send_queued_handshake_initiation(struct wg_peer *peer,
+						bool is_retry);
+void wg_packet_send_handshake_response(struct wg_peer *peer);
+void wg_packet_send_handshake_cookie(struct wg_device *wg,
+				     struct sk_buff *initiating_skb,
+				     __le32 sender_index);
+void wg_packet_send_keepalive(struct wg_peer *peer);
+void wg_packet_purge_staged_packets(struct wg_peer *peer);
+void wg_packet_send_staged_packets(struct wg_peer *peer);
+/* Workqueue workers: */
+void wg_packet_handshake_send_worker(struct work_struct *work);
+void wg_packet_tx_worker(struct work_struct *work);
+void wg_packet_encrypt_worker(struct work_struct *work);
+
+enum packet_state {
+	PACKET_STATE_UNCRYPTED,
+	PACKET_STATE_CRYPTED,
+	PACKET_STATE_DEAD
+};
+
+struct packet_cb {
+	u64 nonce;
+	struct noise_keypair *keypair;
+	atomic_t state;
+	u32 mtu;
+	u8 ds;
+};
+
+#define PACKET_CB(skb) ((struct packet_cb *)((skb)->cb))
+#define PACKET_PEER(skb) (PACKET_CB(skb)->keypair->entry.peer)
+
+static inline bool wg_check_packet_protocol(struct sk_buff *skb)
+{
+	__be16 real_protocol = ip_tunnel_parse_protocol(skb);
+	return real_protocol && skb->protocol == real_protocol;
+}
+
+static inline void wg_reset_packet(struct sk_buff *skb, bool encapsulating)
+{
+	u8 l4_hash = skb->l4_hash;
+	u8 sw_hash = skb->sw_hash;
+	u32 hash = skb->hash;
+	skb_scrub_packet(skb, true);
+	memset(&skb->headers, 0, sizeof(skb->headers));
+	if (encapsulating) {
+		skb->l4_hash = l4_hash;
+		skb->sw_hash = sw_hash;
+		skb->hash = hash;
+	}
+	skb->queue_mapping = 0;
+	skb->nohdr = 0;
+	skb->peeked = 0;
+	skb->mac_len = 0;
+	skb->dev = NULL;
+#ifdef CONFIG_NET_SCHED
+	skb->tc_index = 0;
+#endif
+	skb_reset_redirect(skb);
+	skb->hdr_len = skb_headroom(skb);
+	skb_reset_mac_header(skb);
+	skb_reset_network_header(skb);
+	skb_reset_transport_header(skb);
+	skb_probe_transport_header(skb);
+	skb_reset_inner_headers(skb);
+}
+
+static inline int wg_cpumask_choose_online(int *stored_cpu, unsigned int id)
+{
+	unsigned int cpu = *stored_cpu;
+
+	while (unlikely(cpu >= nr_cpu_ids || !cpu_online(cpu)))
+		cpu = *stored_cpu = cpumask_nth(id % num_online_cpus(), cpu_online_mask);
+
+	return cpu;
+}
+
+/* This function is racy, in the sense that it's called while last_cpu is
+ * unlocked, so it could return the same CPU twice. Adding locking or using
+ * atomic sequence numbers is slower though, and the consequences of racing are
+ * harmless, so live with it.
+ */
+static inline int wg_cpumask_next_online(int *last_cpu)
+{
+	int cpu = cpumask_next(READ_ONCE(*last_cpu), cpu_online_mask);
+	if (cpu >= nr_cpu_ids)
+		cpu = cpumask_first(cpu_online_mask);
+	WRITE_ONCE(*last_cpu, cpu);
+	return cpu;
+}
+
+void wg_prev_queue_init(struct prev_queue *queue);
+
+/* Multi producer */
+bool wg_prev_queue_enqueue(struct prev_queue *queue, struct sk_buff *skb);
+
+/* Single consumer */
+struct sk_buff *wg_prev_queue_dequeue(struct prev_queue *queue);
+
+/* Single consumer */
+static inline struct sk_buff *wg_prev_queue_peek(struct prev_queue *queue)
+{
+	if (queue->peeked)
+		return queue->peeked;
+	queue->peeked = wg_prev_queue_dequeue(queue);
+	return queue->peeked;
+}
+
+/* Single consumer */
+static inline void wg_prev_queue_drop_peeked(struct prev_queue *queue)
+{
+	queue->peeked = NULL;
+}
+
+static inline int wg_queue_enqueue_per_device_and_peer(
+	struct crypt_queue *device_queue, struct prev_queue *peer_queue,
+	struct sk_buff *skb, struct workqueue_struct *wq)
+{
+	int cpu;
+
+	atomic_set_release(&PACKET_CB(skb)->state, PACKET_STATE_UNCRYPTED);
+	/* We first queue this up for the peer ingestion, but the consumer
+	 * will wait for the state to change to CRYPTED or DEAD before.
+	 */
+	if (unlikely(!wg_prev_queue_enqueue(peer_queue, skb)))
+		return -ENOSPC;
+
+	/* Then we queue it up in the device queue, which consumes the
+	 * packet as soon as it can.
+	 */
+	cpu = wg_cpumask_next_online(&device_queue->last_cpu);
+	if (unlikely(ptr_ring_produce_bh(&device_queue->ring, skb)))
+		return -EPIPE;
+	queue_work_on(cpu, wq, &per_cpu_ptr(device_queue->worker, cpu)->work);
+	return 0;
+}
+
+static inline void wg_queue_enqueue_per_peer_tx(struct sk_buff *skb, enum packet_state state)
+{
+	/* We take a reference, because as soon as we call atomic_set, the
+	 * peer can be freed from below us.
+	 */
+	struct wg_peer *peer = wg_peer_get(PACKET_PEER(skb));
+
+	atomic_set_release(&PACKET_CB(skb)->state, state);
+	queue_work_on(wg_cpumask_choose_online(&peer->serial_work_cpu, peer->internal_id),
+		      peer->device->packet_crypt_wq, &peer->transmit_packet_work);
+	wg_peer_put(peer);
+}
+
+static inline void wg_queue_enqueue_per_peer_rx(struct sk_buff *skb, enum packet_state state)
+{
+	/* We take a reference, because as soon as we call atomic_set, the
+	 * peer can be freed from below us.
+	 */
+	struct wg_peer *peer = wg_peer_get(PACKET_PEER(skb));
+
+	atomic_set_release(&PACKET_CB(skb)->state, state);
+	napi_schedule(&peer->napi);
+	wg_peer_put(peer);
+}
+
+#ifdef DEBUG
+bool wg_packet_counter_selftest(void);
+#endif
+
+#endif /* _WG_QUEUEING_H */