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, 144 insertions, 0 deletions

diff --git a/drivers/misc/smpro-misc.c b/drivers/misc/smpro-misc.c
new file mode 100644
index 000000000..2ca5e3bcc
--- /dev/null
+++ b/drivers/misc/smpro-misc.c
@@ -0,0 +1,144 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Ampere Computing SoC's SMpro Misc Driver
+ *
+ * Copyright (c) 2022, Ampere Computing LLC
+ */
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/regmap.h>
+
+/* Boot Stage/Progress Registers */
+#define BOOTSTAGE	0xB0
+#define BOOTSTAGE_LO	0xB1
+#define CUR_BOOTSTAGE	0xB2
+#define BOOTSTAGE_HI	0xB3
+
+/* SOC State Registers */
+#define SOC_POWER_LIMIT		0xE5
+
+struct smpro_misc {
+	struct regmap *regmap;
+};
+
+static ssize_t boot_progress_show(struct device *dev, struct device_attribute *da, char *buf)
+{
+	struct smpro_misc *misc = dev_get_drvdata(dev);
+	u16 boot_progress[3] = { 0 };
+	u32 bootstage;
+	u8 boot_stage;
+	u8 cur_stage;
+	u32 reg_lo;
+	u32 reg;
+	int ret;
+
+	/* Read current boot stage */
+	ret = regmap_read(misc->regmap, CUR_BOOTSTAGE, &reg);
+	if (ret)
+		return ret;
+
+	cur_stage = reg & 0xff;
+
+	ret = regmap_read(misc->regmap, BOOTSTAGE, &bootstage);
+	if (ret)
+		return ret;
+
+	boot_stage = (bootstage >> 8) & 0xff;
+
+	if (boot_stage > cur_stage)
+		return -EINVAL;
+
+	ret = regmap_read(misc->regmap,	BOOTSTAGE_LO, &reg_lo);
+	if (!ret)
+		ret = regmap_read(misc->regmap, BOOTSTAGE_HI, &reg);
+	if (ret)
+		return ret;
+
+	/* Firmware to report new boot stage next time */
+	if (boot_stage < cur_stage) {
+		ret = regmap_write(misc->regmap, BOOTSTAGE, ((bootstage & 0xff00) | 0x1));
+		if (ret)
+			return ret;
+	}
+
+	boot_progress[0] = bootstage;
+	boot_progress[1] = swab16(reg);
+	boot_progress[2] = swab16(reg_lo);
+
+	return sysfs_emit(buf, "%*phN\n", (int)sizeof(boot_progress), boot_progress);
+}
+
+static DEVICE_ATTR_RO(boot_progress);
+
+static ssize_t soc_power_limit_show(struct device *dev, struct device_attribute *da, char *buf)
+{
+	struct smpro_misc *misc = dev_get_drvdata(dev);
+	unsigned int value;
+	int ret;
+
+	ret = regmap_read(misc->regmap, SOC_POWER_LIMIT, &value);
+	if (ret)
+		return ret;
+
+	return sysfs_emit(buf, "%d\n", value);
+}
+
+static ssize_t soc_power_limit_store(struct device *dev, struct device_attribute *da,
+				     const char *buf, size_t count)
+{
+	struct smpro_misc *misc = dev_get_drvdata(dev);
+	unsigned long val;
+	s32 ret;
+
+	ret = kstrtoul(buf, 0, &val);
+	if (ret)
+		return ret;
+
+	ret = regmap_write(misc->regmap, SOC_POWER_LIMIT, (unsigned int)val);
+	if (ret)
+		return -EPROTO;
+
+	return count;
+}
+
+static DEVICE_ATTR_RW(soc_power_limit);
+
+static struct attribute *smpro_misc_attrs[] = {
+	&dev_attr_boot_progress.attr,
+	&dev_attr_soc_power_limit.attr,
+	NULL
+};
+
+ATTRIBUTE_GROUPS(smpro_misc);
+
+static int smpro_misc_probe(struct platform_device *pdev)
+{
+	struct smpro_misc *misc;
+
+	misc = devm_kzalloc(&pdev->dev, sizeof(struct smpro_misc), GFP_KERNEL);
+	if (!misc)
+		return -ENOMEM;
+
+	platform_set_drvdata(pdev, misc);
+
+	misc->regmap = dev_get_regmap(pdev->dev.parent, NULL);
+	if (!misc->regmap)
+		return -ENODEV;
+
+	return 0;
+}
+
+static struct platform_driver smpro_misc_driver = {
+	.probe		= smpro_misc_probe,
+	.driver = {
+		.name	= "smpro-misc",
+		.dev_groups = smpro_misc_groups,
+	},
+};
+
+module_platform_driver(smpro_misc_driver);
+
+MODULE_AUTHOR("Tung Nguyen <tungnguyen@os.amperecomputing.com>");
+MODULE_AUTHOR("Quan Nguyen <quan@os.amperecomputing.com>");
+MODULE_DESCRIPTION("Ampere Altra SMpro Misc driver");
+MODULE_LICENSE("GPL");