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, 127 insertions, 0 deletions

diff --git a/drivers/reset/reset-ath79.c b/drivers/reset/reset-ath79.c
new file mode 100644
index 000000000..1a2c0f8a5
--- /dev/null
+++ b/drivers/reset/reset-ath79.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * AR71xx Reset Controller Driver
+ * Author: Alban Bedel
+ *
+ * Copyright (C) 2015 Alban Bedel <albeu@free.fr>
+ */
+
+#include <linux/io.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+#include <linux/reset-controller.h>
+#include <linux/reboot.h>
+
+struct ath79_reset {
+	struct reset_controller_dev rcdev;
+	void __iomem *base;
+	spinlock_t lock;
+};
+
+#define FULL_CHIP_RESET 24
+
+static int ath79_reset_update(struct reset_controller_dev *rcdev,
+			unsigned long id, bool assert)
+{
+	struct ath79_reset *ath79_reset =
+		container_of(rcdev, struct ath79_reset, rcdev);
+	unsigned long flags;
+	u32 val;
+
+	spin_lock_irqsave(&ath79_reset->lock, flags);
+	val = readl(ath79_reset->base);
+	if (assert)
+		val |= BIT(id);
+	else
+		val &= ~BIT(id);
+	writel(val, ath79_reset->base);
+	spin_unlock_irqrestore(&ath79_reset->lock, flags);
+
+	return 0;
+}
+
+static int ath79_reset_assert(struct reset_controller_dev *rcdev,
+			unsigned long id)
+{
+	return ath79_reset_update(rcdev, id, true);
+}
+
+static int ath79_reset_deassert(struct reset_controller_dev *rcdev,
+				unsigned long id)
+{
+	return ath79_reset_update(rcdev, id, false);
+}
+
+static int ath79_reset_status(struct reset_controller_dev *rcdev,
+			unsigned long id)
+{
+	struct ath79_reset *ath79_reset =
+		container_of(rcdev, struct ath79_reset, rcdev);
+	u32 val;
+
+	val = readl(ath79_reset->base);
+
+	return !!(val & BIT(id));
+}
+
+static const struct reset_control_ops ath79_reset_ops = {
+	.assert = ath79_reset_assert,
+	.deassert = ath79_reset_deassert,
+	.status = ath79_reset_status,
+};
+
+static int ath79_reset_restart_handler(struct sys_off_data *data)
+{
+	struct ath79_reset *ath79_reset = data->cb_data;
+
+	ath79_reset_assert(&ath79_reset->rcdev, FULL_CHIP_RESET);
+
+	return NOTIFY_DONE;
+}
+
+static int ath79_reset_probe(struct platform_device *pdev)
+{
+	struct ath79_reset *ath79_reset;
+	int err;
+
+	ath79_reset = devm_kzalloc(&pdev->dev,
+				sizeof(*ath79_reset), GFP_KERNEL);
+	if (!ath79_reset)
+		return -ENOMEM;
+
+	ath79_reset->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(ath79_reset->base))
+		return PTR_ERR(ath79_reset->base);
+
+	spin_lock_init(&ath79_reset->lock);
+	ath79_reset->rcdev.ops = &ath79_reset_ops;
+	ath79_reset->rcdev.owner = THIS_MODULE;
+	ath79_reset->rcdev.of_node = pdev->dev.of_node;
+	ath79_reset->rcdev.of_reset_n_cells = 1;
+	ath79_reset->rcdev.nr_resets = 32;
+
+	err = devm_reset_controller_register(&pdev->dev, &ath79_reset->rcdev);
+	if (err)
+		return err;
+
+	err = devm_register_restart_handler(&pdev->dev, ath79_reset_restart_handler, ath79_reset);
+	if (err)
+		dev_warn(&pdev->dev, "Failed to register restart handler\n");
+
+	return 0;
+}
+
+static const struct of_device_id ath79_reset_dt_ids[] = {
+	{ .compatible = "qca,ar7100-reset", },
+	{ },
+};
+
+static struct platform_driver ath79_reset_driver = {
+	.probe	= ath79_reset_probe,
+	.driver = {
+		.name			= "ath79-reset",
+		.of_match_table		= ath79_reset_dt_ids,
+		.suppress_bind_attrs	= true,
+	},
+};
+builtin_platform_driver(ath79_reset_driver);