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, 206 insertions, 0 deletions

diff --git a/net/lapb/lapb_timer.c b/net/lapb/lapb_timer.c
new file mode 100644
index 000000000..9fde6cf20
--- /dev/null
+++ b/net/lapb/lapb_timer.c
@@ -0,0 +1,206 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ *	LAPB release 002
+ *
+ *	This code REQUIRES 2.1.15 or higher/ NET3.038
+ *
+ *	History
+ *	LAPB 001	Jonathan Naylor	Started Coding
+ *	LAPB 002	Jonathan Naylor	New timer architecture.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/in.h>
+#include <linux/kernel.h>
+#include <linux/jiffies.h>
+#include <linux/timer.h>
+#include <linux/string.h>
+#include <linux/sockios.h>
+#include <linux/net.h>
+#include <linux/inet.h>
+#include <linux/skbuff.h>
+#include <net/sock.h>
+#include <linux/uaccess.h>
+#include <linux/fcntl.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <net/lapb.h>
+
+static void lapb_t1timer_expiry(struct timer_list *);
+static void lapb_t2timer_expiry(struct timer_list *);
+
+void lapb_start_t1timer(struct lapb_cb *lapb)
+{
+	timer_delete(&lapb->t1timer);
+
+	lapb->t1timer.function = lapb_t1timer_expiry;
+	lapb->t1timer.expires  = jiffies + lapb->t1;
+
+	lapb->t1timer_running = true;
+	add_timer(&lapb->t1timer);
+}
+
+void lapb_start_t2timer(struct lapb_cb *lapb)
+{
+	timer_delete(&lapb->t2timer);
+
+	lapb->t2timer.function = lapb_t2timer_expiry;
+	lapb->t2timer.expires  = jiffies + lapb->t2;
+
+	lapb->t2timer_running = true;
+	add_timer(&lapb->t2timer);
+}
+
+void lapb_stop_t1timer(struct lapb_cb *lapb)
+{
+	lapb->t1timer_running = false;
+	timer_delete(&lapb->t1timer);
+}
+
+void lapb_stop_t2timer(struct lapb_cb *lapb)
+{
+	lapb->t2timer_running = false;
+	timer_delete(&lapb->t2timer);
+}
+
+int lapb_t1timer_running(struct lapb_cb *lapb)
+{
+	return lapb->t1timer_running;
+}
+
+static void lapb_t2timer_expiry(struct timer_list *t)
+{
+	struct lapb_cb *lapb = timer_container_of(lapb, t, t2timer);
+
+	spin_lock_bh(&lapb->lock);
+	if (timer_pending(&lapb->t2timer)) /* A new timer has been set up */
+		goto out;
+	if (!lapb->t2timer_running) /* The timer has been stopped */
+		goto out;
+
+	if (lapb->condition & LAPB_ACK_PENDING_CONDITION) {
+		lapb->condition &= ~LAPB_ACK_PENDING_CONDITION;
+		lapb_timeout_response(lapb);
+	}
+	lapb->t2timer_running = false;
+
+out:
+	spin_unlock_bh(&lapb->lock);
+}
+
+static void lapb_t1timer_expiry(struct timer_list *t)
+{
+	struct lapb_cb *lapb = timer_container_of(lapb, t, t1timer);
+
+	spin_lock_bh(&lapb->lock);
+	if (timer_pending(&lapb->t1timer)) /* A new timer has been set up */
+		goto out;
+	if (!lapb->t1timer_running) /* The timer has been stopped */
+		goto out;
+
+	switch (lapb->state) {
+
+		/*
+		 *	If we are a DCE, send DM up to N2 times, then switch to
+		 *	STATE_1 and send SABM(E).
+		 */
+		case LAPB_STATE_0:
+			if (lapb->mode & LAPB_DCE &&
+			    lapb->n2count != lapb->n2) {
+				lapb->n2count++;
+				lapb_send_control(lapb, LAPB_DM, LAPB_POLLOFF, LAPB_RESPONSE);
+			} else {
+				lapb->state = LAPB_STATE_1;
+				lapb_establish_data_link(lapb);
+			}
+			break;
+
+		/*
+		 *	Awaiting connection state, send SABM(E), up to N2 times.
+		 */
+		case LAPB_STATE_1:
+			if (lapb->n2count == lapb->n2) {
+				lapb_clear_queues(lapb);
+				lapb->state = LAPB_STATE_0;
+				lapb_disconnect_indication(lapb, LAPB_TIMEDOUT);
+				lapb_dbg(0, "(%p) S1 -> S0\n", lapb->dev);
+				lapb->t1timer_running = false;
+				goto out;
+			} else {
+				lapb->n2count++;
+				if (lapb->mode & LAPB_EXTENDED) {
+					lapb_dbg(1, "(%p) S1 TX SABME(1)\n",
+						 lapb->dev);
+					lapb_send_control(lapb, LAPB_SABME, LAPB_POLLON, LAPB_COMMAND);
+				} else {
+					lapb_dbg(1, "(%p) S1 TX SABM(1)\n",
+						 lapb->dev);
+					lapb_send_control(lapb, LAPB_SABM, LAPB_POLLON, LAPB_COMMAND);
+				}
+			}
+			break;
+
+		/*
+		 *	Awaiting disconnection state, send DISC, up to N2 times.
+		 */
+		case LAPB_STATE_2:
+			if (lapb->n2count == lapb->n2) {
+				lapb_clear_queues(lapb);
+				lapb->state = LAPB_STATE_0;
+				lapb_disconnect_confirmation(lapb, LAPB_TIMEDOUT);
+				lapb_dbg(0, "(%p) S2 -> S0\n", lapb->dev);
+				lapb->t1timer_running = false;
+				goto out;
+			} else {
+				lapb->n2count++;
+				lapb_dbg(1, "(%p) S2 TX DISC(1)\n", lapb->dev);
+				lapb_send_control(lapb, LAPB_DISC, LAPB_POLLON, LAPB_COMMAND);
+			}
+			break;
+
+		/*
+		 *	Data transfer state, restransmit I frames, up to N2 times.
+		 */
+		case LAPB_STATE_3:
+			if (lapb->n2count == lapb->n2) {
+				lapb_clear_queues(lapb);
+				lapb->state = LAPB_STATE_0;
+				lapb_stop_t2timer(lapb);
+				lapb_disconnect_indication(lapb, LAPB_TIMEDOUT);
+				lapb_dbg(0, "(%p) S3 -> S0\n", lapb->dev);
+				lapb->t1timer_running = false;
+				goto out;
+			} else {
+				lapb->n2count++;
+				lapb_requeue_frames(lapb);
+				lapb_kick(lapb);
+			}
+			break;
+
+		/*
+		 *	Frame reject state, restransmit FRMR frames, up to N2 times.
+		 */
+		case LAPB_STATE_4:
+			if (lapb->n2count == lapb->n2) {
+				lapb_clear_queues(lapb);
+				lapb->state = LAPB_STATE_0;
+				lapb_disconnect_indication(lapb, LAPB_TIMEDOUT);
+				lapb_dbg(0, "(%p) S4 -> S0\n", lapb->dev);
+				lapb->t1timer_running = false;
+				goto out;
+			} else {
+				lapb->n2count++;
+				lapb_transmit_frmr(lapb);
+			}
+			break;
+	}
+
+	lapb_start_t1timer(lapb);
+
+out:
+	spin_unlock_bh(&lapb->lock);
+}