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, 194 insertions, 0 deletions

diff --git a/net/x25/x25_dev.c b/net/x25/x25_dev.c
new file mode 100644
index 000000000..fb8ac1aa5
--- /dev/null
+++ b/net/x25/x25_dev.c
@@ -0,0 +1,194 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *	X.25 Packet Layer release 002
+ *
+ *	This is ALPHA test software. This code may break your machine, randomly fail to work with new
+ *	releases, misbehave and/or generally screw up. It might even work.
+ *
+ *	This code REQUIRES 2.1.15 or higher
+ *
+ *	History
+ *	X.25 001	Jonathan Naylor	Started coding.
+ *      2000-09-04	Henner Eisen	Prevent freeing a dangling skb.
+ */
+
+#define pr_fmt(fmt) "X25: " fmt
+
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/skbuff.h>
+#include <linux/slab.h>
+#include <net/sock.h>
+#include <linux/if_arp.h>
+#include <net/x25.h>
+#include <net/x25device.h>
+
+static int x25_receive_data(struct sk_buff *skb, struct x25_neigh *nb)
+{
+	struct sock *sk;
+	unsigned short frametype;
+	unsigned int lci;
+
+	if (!pskb_may_pull(skb, X25_STD_MIN_LEN))
+		return 0;
+
+	frametype = skb->data[2];
+	lci = ((skb->data[0] << 8) & 0xF00) + ((skb->data[1] << 0) & 0x0FF);
+
+	/*
+	 *	LCI of zero is always for us, and its always a link control
+	 *	frame.
+	 */
+	if (lci == 0) {
+		x25_link_control(skb, nb, frametype);
+		return 0;
+	}
+
+	/*
+	 *	Find an existing socket.
+	 */
+	if ((sk = x25_find_socket(lci, nb)) != NULL) {
+		int queued = 1;
+
+		skb_reset_transport_header(skb);
+		bh_lock_sock(sk);
+		if (!sock_owned_by_user(sk)) {
+			queued = x25_process_rx_frame(sk, skb);
+		} else {
+			queued = !sk_add_backlog(sk, skb, READ_ONCE(sk->sk_rcvbuf));
+		}
+		bh_unlock_sock(sk);
+		sock_put(sk);
+		return queued;
+	}
+
+	/*
+	 *	Is is a Call Request ? if so process it.
+	 */
+	if (frametype == X25_CALL_REQUEST)
+		return x25_rx_call_request(skb, nb, lci);
+
+	/*
+	 * 	Its not a Call Request, nor is it a control frame.
+	 *	Can we forward it?
+	 */
+
+	if (x25_forward_data(lci, nb, skb)) {
+		if (frametype == X25_CLEAR_CONFIRMATION) {
+			x25_clear_forward_by_lci(lci);
+		}
+		kfree_skb(skb);
+		return 1;
+	}
+
+/*
+	x25_transmit_clear_request(nb, lci, 0x0D);
+*/
+
+	if (frametype != X25_CLEAR_CONFIRMATION)
+		pr_debug("x25_receive_data(): unknown frame type %2x\n",frametype);
+
+	return 0;
+}
+
+int x25_lapb_receive_frame(struct sk_buff *skb, struct net_device *dev,
+			   struct packet_type *ptype, struct net_device *orig_dev)
+{
+	struct sk_buff *nskb;
+	struct x25_neigh *nb;
+
+	if (!net_eq(dev_net(dev), &init_net))
+		goto drop;
+
+	nskb = skb_copy(skb, GFP_ATOMIC);
+	if (!nskb)
+		goto drop;
+	kfree_skb(skb);
+	skb = nskb;
+
+	/*
+	 * Packet received from unrecognised device, throw it away.
+	 */
+	nb = x25_get_neigh(dev);
+	if (!nb) {
+		pr_debug("unknown neighbour - %s\n", dev->name);
+		goto drop;
+	}
+
+	if (!pskb_may_pull(skb, 1)) {
+		x25_neigh_put(nb);
+		goto drop;
+	}
+
+	switch (skb->data[0]) {
+
+	case X25_IFACE_DATA:
+		skb_pull(skb, 1);
+		if (x25_receive_data(skb, nb)) {
+			x25_neigh_put(nb);
+			goto out;
+		}
+		break;
+
+	case X25_IFACE_CONNECT:
+		x25_link_established(nb);
+		break;
+
+	case X25_IFACE_DISCONNECT:
+		x25_link_terminated(nb);
+		break;
+	}
+	x25_neigh_put(nb);
+drop:
+	kfree_skb(skb);
+out:
+	return 0;
+}
+
+void x25_establish_link(struct x25_neigh *nb)
+{
+	struct sk_buff *skb;
+	unsigned char *ptr;
+
+	switch (nb->dev->type) {
+	case ARPHRD_X25:
+		if ((skb = alloc_skb(1, GFP_ATOMIC)) == NULL) {
+			pr_err("x25_dev: out of memory\n");
+			return;
+		}
+		ptr  = skb_put(skb, 1);
+		*ptr = X25_IFACE_CONNECT;
+		break;
+
+	default:
+		return;
+	}
+
+	skb->protocol = htons(ETH_P_X25);
+	skb->dev      = nb->dev;
+
+	dev_queue_xmit(skb);
+}
+
+void x25_send_frame(struct sk_buff *skb, struct x25_neigh *nb)
+{
+	unsigned char *dptr;
+
+	skb_reset_network_header(skb);
+
+	switch (nb->dev->type) {
+	case ARPHRD_X25:
+		dptr  = skb_push(skb, 1);
+		*dptr = X25_IFACE_DATA;
+		break;
+
+	default:
+		kfree_skb(skb);
+		return;
+	}
+
+	skb->protocol = htons(ETH_P_X25);
+	skb->dev      = nb->dev;
+
+	dev_queue_xmit(skb);
+}