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, 154 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-max6902.c b/drivers/rtc/rtc-max6902.c
new file mode 100644
index 000000000..daaeb6fb6
--- /dev/null
+++ b/drivers/rtc/rtc-max6902.c
@@ -0,0 +1,154 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* drivers/rtc/rtc-max6902.c
+ *
+ * Copyright (C) 2006 8D Technologies inc.
+ * Copyright (C) 2004 Compulab Ltd.
+ *
+ * Driver for MAX6902 spi RTC
+ */
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/init.h>
+#include <linux/rtc.h>
+#include <linux/spi/spi.h>
+#include <linux/bcd.h>
+
+#define MAX6902_REG_SECONDS		0x01
+#define MAX6902_REG_MINUTES		0x03
+#define MAX6902_REG_HOURS		0x05
+#define MAX6902_REG_DATE		0x07
+#define MAX6902_REG_MONTH		0x09
+#define MAX6902_REG_DAY			0x0B
+#define MAX6902_REG_YEAR		0x0D
+#define MAX6902_REG_CONTROL		0x0F
+#define MAX6902_REG_CENTURY		0x13
+
+static int max6902_set_reg(struct device *dev, unsigned char address,
+				unsigned char data)
+{
+	struct spi_device *spi = to_spi_device(dev);
+	unsigned char buf[2];
+
+	/* MSB must be '0' to write */
+	buf[0] = address & 0x7f;
+	buf[1] = data;
+
+	return spi_write_then_read(spi, buf, 2, NULL, 0);
+}
+
+static int max6902_get_reg(struct device *dev, unsigned char address,
+				unsigned char *data)
+{
+	struct spi_device *spi = to_spi_device(dev);
+
+	/* Set MSB to indicate read */
+	*data = address | 0x80;
+
+	return spi_write_then_read(spi, data, 1, data, 1);
+}
+
+static int max6902_read_time(struct device *dev, struct rtc_time *dt)
+{
+	int err, century;
+	struct spi_device *spi = to_spi_device(dev);
+	unsigned char buf[8];
+
+	buf[0] = 0xbf;	/* Burst read */
+
+	err = spi_write_then_read(spi, buf, 1, buf, 8);
+	if (err != 0)
+		return err;
+
+	/* The chip sends data in this order:
+	 * Seconds, Minutes, Hours, Date, Month, Day, Year */
+	dt->tm_sec	= bcd2bin(buf[0]);
+	dt->tm_min	= bcd2bin(buf[1]);
+	dt->tm_hour	= bcd2bin(buf[2]);
+	dt->tm_mday	= bcd2bin(buf[3]);
+	dt->tm_mon	= bcd2bin(buf[4]) - 1;
+	dt->tm_wday	= bcd2bin(buf[5]);
+	dt->tm_year	= bcd2bin(buf[6]);
+
+	/* Read century */
+	err = max6902_get_reg(dev, MAX6902_REG_CENTURY, &buf[0]);
+	if (err != 0)
+		return err;
+
+	century = bcd2bin(buf[0]) * 100;
+
+	dt->tm_year += century;
+	dt->tm_year -= 1900;
+
+	return 0;
+}
+
+static int max6902_set_time(struct device *dev, struct rtc_time *dt)
+{
+	dt->tm_year = dt->tm_year + 1900;
+
+	/* Remove write protection */
+	max6902_set_reg(dev, MAX6902_REG_CONTROL, 0);
+
+	max6902_set_reg(dev, MAX6902_REG_SECONDS, bin2bcd(dt->tm_sec));
+	max6902_set_reg(dev, MAX6902_REG_MINUTES, bin2bcd(dt->tm_min));
+	max6902_set_reg(dev, MAX6902_REG_HOURS, bin2bcd(dt->tm_hour));
+
+	max6902_set_reg(dev, MAX6902_REG_DATE, bin2bcd(dt->tm_mday));
+	max6902_set_reg(dev, MAX6902_REG_MONTH, bin2bcd(dt->tm_mon + 1));
+	max6902_set_reg(dev, MAX6902_REG_DAY, bin2bcd(dt->tm_wday));
+	max6902_set_reg(dev, MAX6902_REG_YEAR, bin2bcd(dt->tm_year % 100));
+	max6902_set_reg(dev, MAX6902_REG_CENTURY, bin2bcd(dt->tm_year / 100));
+
+	/* Compulab used a delay here. However, the datasheet
+	 * does not mention a delay being required anywhere... */
+	/* delay(2000); */
+
+	/* Write protect */
+	max6902_set_reg(dev, MAX6902_REG_CONTROL, 0x80);
+
+	return 0;
+}
+
+static const struct rtc_class_ops max6902_rtc_ops = {
+	.read_time	= max6902_read_time,
+	.set_time	= max6902_set_time,
+};
+
+static int max6902_probe(struct spi_device *spi)
+{
+	struct rtc_device *rtc;
+	unsigned char tmp;
+	int res;
+
+	spi->mode = SPI_MODE_3;
+	spi->bits_per_word = 8;
+	spi_setup(spi);
+
+	res = max6902_get_reg(&spi->dev, MAX6902_REG_SECONDS, &tmp);
+	if (res != 0)
+		return res;
+
+	rtc = devm_rtc_device_register(&spi->dev, "max6902",
+				&max6902_rtc_ops, THIS_MODULE);
+	if (IS_ERR(rtc))
+		return PTR_ERR(rtc);
+
+	spi_set_drvdata(spi, rtc);
+	return 0;
+}
+
+static struct spi_driver max6902_driver = {
+	.driver = {
+		.name	= "rtc-max6902",
+	},
+	.probe	= max6902_probe,
+};
+
+module_spi_driver(max6902_driver);
+
+MODULE_DESCRIPTION("max6902 spi RTC driver");
+MODULE_AUTHOR("Raphael Assenat");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("spi:rtc-max6902");