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, 123 insertions, 0 deletions

diff --git a/drivers/mfd/atmel-flexcom.c b/drivers/mfd/atmel-flexcom.c
new file mode 100644
index 000000000..d5df5466e
--- /dev/null
+++ b/drivers/mfd/atmel-flexcom.c
@@ -0,0 +1,123 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Driver for Atmel Flexcom
+ *
+ * Copyright (C) 2015 Atmel Corporation
+ *
+ * Author: Cyrille Pitchen <cyrille.pitchen@atmel.com>
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/clk.h>
+#include <dt-bindings/mfd/atmel-flexcom.h>
+
+/* I/O register offsets */
+#define FLEX_MR		0x0	/* Mode Register */
+#define FLEX_VERSION	0xfc	/* Version Register */
+
+/* Mode Register bit fields */
+#define FLEX_MR_OPMODE_OFFSET	(0)  /* Operating Mode */
+#define FLEX_MR_OPMODE_MASK	(0x3 << FLEX_MR_OPMODE_OFFSET)
+#define FLEX_MR_OPMODE(opmode)	(((opmode) << FLEX_MR_OPMODE_OFFSET) &	\
+				 FLEX_MR_OPMODE_MASK)
+
+struct atmel_flexcom {
+	void __iomem *base;
+	u32 opmode;
+	struct clk *clk;
+};
+
+static int atmel_flexcom_probe(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct atmel_flexcom *ddata;
+	int err;
+
+	ddata = devm_kzalloc(&pdev->dev, sizeof(*ddata), GFP_KERNEL);
+	if (!ddata)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, ddata);
+
+	err = of_property_read_u32(np, "atmel,flexcom-mode", &ddata->opmode);
+	if (err)
+		return err;
+
+	if (ddata->opmode < ATMEL_FLEXCOM_MODE_USART ||
+	    ddata->opmode > ATMEL_FLEXCOM_MODE_TWI)
+		return -EINVAL;
+
+	ddata->base = devm_platform_get_and_ioremap_resource(pdev, 0, NULL);
+	if (IS_ERR(ddata->base))
+		return PTR_ERR(ddata->base);
+
+	ddata->clk = devm_clk_get(&pdev->dev, NULL);
+	if (IS_ERR(ddata->clk))
+		return PTR_ERR(ddata->clk);
+
+	err = clk_prepare_enable(ddata->clk);
+	if (err)
+		return err;
+
+	/*
+	 * Set the Operating Mode in the Mode Register: only the selected device
+	 * is clocked. Hence, registers of the other serial devices remain
+	 * inaccessible and are read as zero. Also the external I/O lines of the
+	 * Flexcom are muxed to reach the selected device.
+	 */
+	writel(FLEX_MR_OPMODE(ddata->opmode), ddata->base + FLEX_MR);
+
+	clk_disable_unprepare(ddata->clk);
+
+	return devm_of_platform_populate(&pdev->dev);
+}
+
+static const struct of_device_id atmel_flexcom_of_match[] = {
+	{ .compatible = "atmel,sama5d2-flexcom" },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, atmel_flexcom_of_match);
+
+static int __maybe_unused atmel_flexcom_resume_noirq(struct device *dev)
+{
+	struct atmel_flexcom *ddata = dev_get_drvdata(dev);
+	int err;
+	u32 val;
+
+	err = clk_prepare_enable(ddata->clk);
+	if (err)
+		return err;
+
+	val = FLEX_MR_OPMODE(ddata->opmode);
+	writel(val, ddata->base + FLEX_MR);
+
+	clk_disable_unprepare(ddata->clk);
+
+	return 0;
+}
+
+static const struct dev_pm_ops __maybe_unused atmel_flexcom_pm_ops = {
+	.resume_noirq = atmel_flexcom_resume_noirq,
+};
+
+static struct platform_driver atmel_flexcom_driver = {
+	.probe	= atmel_flexcom_probe,
+	.driver	= {
+		.name		= "atmel_flexcom",
+		.pm		= pm_ptr(&atmel_flexcom_pm_ops),
+		.of_match_table	= atmel_flexcom_of_match,
+	},
+};
+
+module_platform_driver(atmel_flexcom_driver);
+
+MODULE_AUTHOR("Cyrille Pitchen <cyrille.pitchen@atmel.com>");
+MODULE_DESCRIPTION("Atmel Flexcom MFD driver");
+MODULE_LICENSE("GPL v2");