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, 126 insertions, 0 deletions

diff --git a/drivers/reset/reset-brcmstb.c b/drivers/reset/reset-brcmstb.c
new file mode 100644
index 000000000..810fe7645
--- /dev/null
+++ b/drivers/reset/reset-brcmstb.c
@@ -0,0 +1,126 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Broadcom STB generic reset controller for SW_INIT style reset controller
+ *
+ * Author: Florian Fainelli <f.fainelli@gmail.com>
+ * Copyright (C) 2018 Broadcom
+ */
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/reset-controller.h>
+#include <linux/types.h>
+
+struct brcmstb_reset {
+	void __iomem *base;
+	struct reset_controller_dev rcdev;
+};
+
+#define SW_INIT_SET		0x00
+#define SW_INIT_CLEAR		0x04
+#define SW_INIT_STATUS		0x08
+
+#define SW_INIT_BIT(id)		BIT((id) & 0x1f)
+#define SW_INIT_BANK(id)	((id) >> 5)
+
+/* A full bank contains extra registers that we are not utilizing but still
+ * qualify as a single bank.
+ */
+#define SW_INIT_BANK_SIZE	0x18
+
+static inline
+struct brcmstb_reset *to_brcmstb(struct reset_controller_dev *rcdev)
+{
+	return container_of(rcdev, struct brcmstb_reset, rcdev);
+}
+
+static int brcmstb_reset_assert(struct reset_controller_dev *rcdev,
+				unsigned long id)
+{
+	unsigned int off = SW_INIT_BANK(id) * SW_INIT_BANK_SIZE;
+	struct brcmstb_reset *priv = to_brcmstb(rcdev);
+
+	writel_relaxed(SW_INIT_BIT(id), priv->base + off + SW_INIT_SET);
+
+	return 0;
+}
+
+static int brcmstb_reset_deassert(struct reset_controller_dev *rcdev,
+				  unsigned long id)
+{
+	unsigned int off = SW_INIT_BANK(id) * SW_INIT_BANK_SIZE;
+	struct brcmstb_reset *priv = to_brcmstb(rcdev);
+
+	writel_relaxed(SW_INIT_BIT(id), priv->base + off + SW_INIT_CLEAR);
+	/* Maximum reset delay after de-asserting a line and seeing block
+	 * operation is typically 14us for the worst case, build some slack
+	 * here.
+	 */
+	usleep_range(100, 200);
+
+	return 0;
+}
+
+static int brcmstb_reset_status(struct reset_controller_dev *rcdev,
+				unsigned long id)
+{
+	unsigned int off = SW_INIT_BANK(id) * SW_INIT_BANK_SIZE;
+	struct brcmstb_reset *priv = to_brcmstb(rcdev);
+
+	return readl_relaxed(priv->base + off + SW_INIT_STATUS) &
+			     SW_INIT_BIT(id);
+}
+
+static const struct reset_control_ops brcmstb_reset_ops = {
+	.assert	= brcmstb_reset_assert,
+	.deassert = brcmstb_reset_deassert,
+	.status = brcmstb_reset_status,
+};
+
+static int brcmstb_reset_probe(struct platform_device *pdev)
+{
+	struct device *kdev = &pdev->dev;
+	struct brcmstb_reset *priv;
+	struct resource *res;
+
+	priv = devm_kzalloc(kdev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);
+
+	dev_set_drvdata(kdev, priv);
+
+	priv->rcdev.owner = THIS_MODULE;
+	priv->rcdev.nr_resets = DIV_ROUND_DOWN_ULL(resource_size(res),
+						   SW_INIT_BANK_SIZE) * 32;
+	priv->rcdev.ops = &brcmstb_reset_ops;
+	priv->rcdev.of_node = kdev->of_node;
+	/* Use defaults: 1 cell and simple xlate function */
+
+	return devm_reset_controller_register(kdev, &priv->rcdev);
+}
+
+static const struct of_device_id brcmstb_reset_of_match[] = {
+	{ .compatible = "brcm,brcmstb-reset" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, brcmstb_reset_of_match);
+
+static struct platform_driver brcmstb_reset_driver = {
+	.probe	= brcmstb_reset_probe,
+	.driver	= {
+		.name = "brcmstb-reset",
+		.of_match_table = brcmstb_reset_of_match,
+	},
+};
+module_platform_driver(brcmstb_reset_driver);
+
+MODULE_AUTHOR("Broadcom");
+MODULE_DESCRIPTION("Broadcom STB reset controller");
+MODULE_LICENSE("GPL");