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, 213 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-ds1302.c b/drivers/rtc/rtc-ds1302.c
new file mode 100644
index 000000000..ecc7d0307
--- /dev/null
+++ b/drivers/rtc/rtc-ds1302.c
@@ -0,0 +1,213 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Dallas DS1302 RTC Support
+ *
+ *  Copyright (C) 2002 David McCullough
+ *  Copyright (C) 2003 - 2007 Paul Mundt
+ */
+
+#include <linux/bcd.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/rtc.h>
+#include <linux/spi/spi.h>
+
+#define	RTC_CMD_READ	0x81		/* Read command */
+#define	RTC_CMD_WRITE	0x80		/* Write command */
+
+#define	RTC_CMD_WRITE_ENABLE	0x00		/* Write enable */
+#define	RTC_CMD_WRITE_DISABLE	0x80		/* Write disable */
+
+#define RTC_ADDR_RAM0	0x20		/* Address of RAM0 */
+#define RTC_ADDR_TCR	0x08		/* Address of trickle charge register */
+#define RTC_CLCK_BURST	0x1F		/* Address of clock burst */
+#define	RTC_CLCK_LEN	0x08		/* Size of clock burst */
+#define	RTC_ADDR_CTRL	0x07		/* Address of control register */
+#define	RTC_ADDR_YEAR	0x06		/* Address of year register */
+#define	RTC_ADDR_DAY	0x05		/* Address of day of week register */
+#define	RTC_ADDR_MON	0x04		/* Address of month register */
+#define	RTC_ADDR_DATE	0x03		/* Address of day of month register */
+#define	RTC_ADDR_HOUR	0x02		/* Address of hour register */
+#define	RTC_ADDR_MIN	0x01		/* Address of minute register */
+#define	RTC_ADDR_SEC	0x00		/* Address of second register */
+
+static int ds1302_rtc_set_time(struct device *dev, struct rtc_time *time)
+{
+	struct spi_device	*spi = dev_get_drvdata(dev);
+	u8		buf[1 + RTC_CLCK_LEN];
+	u8		*bp;
+	int		status;
+
+	/* Enable writing */
+	bp = buf;
+	*bp++ = RTC_ADDR_CTRL << 1 | RTC_CMD_WRITE;
+	*bp++ = RTC_CMD_WRITE_ENABLE;
+
+	status = spi_write_then_read(spi, buf, 2,
+			NULL, 0);
+	if (status)
+		return status;
+
+	/* Write registers starting at the first time/date address. */
+	bp = buf;
+	*bp++ = RTC_CLCK_BURST << 1 | RTC_CMD_WRITE;
+
+	*bp++ = bin2bcd(time->tm_sec);
+	*bp++ = bin2bcd(time->tm_min);
+	*bp++ = bin2bcd(time->tm_hour);
+	*bp++ = bin2bcd(time->tm_mday);
+	*bp++ = bin2bcd(time->tm_mon + 1);
+	*bp++ = time->tm_wday + 1;
+	*bp++ = bin2bcd(time->tm_year % 100);
+	*bp++ = RTC_CMD_WRITE_DISABLE;
+
+	/* use write-then-read since dma from stack is nonportable */
+	return spi_write_then_read(spi, buf, sizeof(buf),
+			NULL, 0);
+}
+
+static int ds1302_rtc_get_time(struct device *dev, struct rtc_time *time)
+{
+	struct spi_device	*spi = dev_get_drvdata(dev);
+	u8		addr = RTC_CLCK_BURST << 1 | RTC_CMD_READ;
+	u8		buf[RTC_CLCK_LEN - 1];
+	int		status;
+
+	/* Use write-then-read to get all the date/time registers
+	 * since dma from stack is nonportable
+	 */
+	status = spi_write_then_read(spi, &addr, sizeof(addr),
+			buf, sizeof(buf));
+	if (status < 0)
+		return status;
+
+	/* Decode the registers */
+	time->tm_sec = bcd2bin(buf[RTC_ADDR_SEC]);
+	time->tm_min = bcd2bin(buf[RTC_ADDR_MIN]);
+	time->tm_hour = bcd2bin(buf[RTC_ADDR_HOUR]);
+	time->tm_wday = buf[RTC_ADDR_DAY] - 1;
+	time->tm_mday = bcd2bin(buf[RTC_ADDR_DATE]);
+	time->tm_mon = bcd2bin(buf[RTC_ADDR_MON]) - 1;
+	time->tm_year = bcd2bin(buf[RTC_ADDR_YEAR]) + 100;
+
+	return 0;
+}
+
+static const struct rtc_class_ops ds1302_rtc_ops = {
+	.read_time	= ds1302_rtc_get_time,
+	.set_time	= ds1302_rtc_set_time,
+};
+
+static int ds1302_probe(struct spi_device *spi)
+{
+	struct rtc_device	*rtc;
+	u8		addr;
+	u8		buf[4];
+	u8		*bp;
+	int		status;
+
+	/* Sanity check board setup data.  This may be hooked up
+	 * in 3wire mode, but we don't care.  Note that unless
+	 * there's an inverter in place, this needs SPI_CS_HIGH!
+	 */
+	if (spi->bits_per_word && (spi->bits_per_word != 8)) {
+		dev_err(&spi->dev, "bad word length\n");
+		return -EINVAL;
+	} else if (spi->max_speed_hz > 2000000) {
+		dev_err(&spi->dev, "speed is too high\n");
+		return -EINVAL;
+	} else if (spi->mode & SPI_CPHA) {
+		dev_err(&spi->dev, "bad mode\n");
+		return -EINVAL;
+	}
+
+	addr = RTC_ADDR_CTRL << 1 | RTC_CMD_READ;
+	status = spi_write_then_read(spi, &addr, sizeof(addr), buf, 1);
+	if (status < 0) {
+		dev_err(&spi->dev, "control register read error %d\n",
+				status);
+		return status;
+	}
+
+	if ((buf[0] & ~RTC_CMD_WRITE_DISABLE) != 0) {
+		status = spi_write_then_read(spi, &addr, sizeof(addr), buf, 1);
+		if (status < 0) {
+			dev_err(&spi->dev, "control register read error %d\n",
+					status);
+			return status;
+		}
+
+		if ((buf[0] & ~RTC_CMD_WRITE_DISABLE) != 0) {
+			dev_err(&spi->dev, "junk in control register\n");
+			return -ENODEV;
+		}
+	}
+	if (buf[0] == 0) {
+		bp = buf;
+		*bp++ = RTC_ADDR_CTRL << 1 | RTC_CMD_WRITE;
+		*bp++ = RTC_CMD_WRITE_DISABLE;
+
+		status = spi_write_then_read(spi, buf, 2, NULL, 0);
+		if (status < 0) {
+			dev_err(&spi->dev, "control register write error %d\n",
+					status);
+			return status;
+		}
+
+		addr = RTC_ADDR_CTRL << 1 | RTC_CMD_READ;
+		status = spi_write_then_read(spi, &addr, sizeof(addr), buf, 1);
+		if (status < 0) {
+			dev_err(&spi->dev,
+					"error %d reading control register\n",
+					status);
+			return status;
+		}
+
+		if (buf[0] != RTC_CMD_WRITE_DISABLE) {
+			dev_err(&spi->dev, "failed to detect chip\n");
+			return -ENODEV;
+		}
+	}
+
+	spi_set_drvdata(spi, spi);
+
+	rtc = devm_rtc_device_register(&spi->dev, "ds1302",
+			&ds1302_rtc_ops, THIS_MODULE);
+	if (IS_ERR(rtc)) {
+		status = PTR_ERR(rtc);
+		dev_err(&spi->dev, "error %d registering rtc\n", status);
+		return status;
+	}
+
+	return 0;
+}
+
+#ifdef CONFIG_OF
+static const struct of_device_id ds1302_dt_ids[] = {
+	{ .compatible = "maxim,ds1302", },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(of, ds1302_dt_ids);
+#endif
+
+static const struct spi_device_id ds1302_spi_ids[] = {
+	{ .name = "ds1302", },
+	{ /* sentinel */ }
+};
+MODULE_DEVICE_TABLE(spi, ds1302_spi_ids);
+
+static struct spi_driver ds1302_driver = {
+	.driver.name	= "rtc-ds1302",
+	.driver.of_match_table = of_match_ptr(ds1302_dt_ids),
+	.probe		= ds1302_probe,
+	.id_table	= ds1302_spi_ids,
+};
+
+module_spi_driver(ds1302_driver);
+
+MODULE_DESCRIPTION("Dallas DS1302 RTC driver");
+MODULE_AUTHOR("Paul Mundt, David McCullough");
+MODULE_LICENSE("GPL v2");