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, 127 insertions, 0 deletions

diff --git a/drivers/pinctrl/mediatek/mtk-eint.h b/drivers/pinctrl/mediatek/mtk-eint.h
new file mode 100644
index 000000000..3cdd6f631
--- /dev/null
+++ b/drivers/pinctrl/mediatek/mtk-eint.h
@@ -0,0 +1,127 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2014-2025 MediaTek Inc.
+ *
+ * Author: Maoguang Meng <maoguang.meng@mediatek.com>
+ *	   Sean Wang <sean.wang@mediatek.com>
+ *	   Hao Chang <ot_chhao.chang@mediatek.com>
+ *	   Qingliang Li <qingliang.li@mediatek.com>
+ */
+#ifndef __MTK_EINT_H
+#define __MTK_EINT_H
+
+#include <linux/irqdomain.h>
+
+struct mtk_eint_regs {
+	unsigned int	stat;
+	unsigned int	ack;
+	unsigned int	mask;
+	unsigned int	mask_set;
+	unsigned int	mask_clr;
+	unsigned int	sens;
+	unsigned int	sens_set;
+	unsigned int	sens_clr;
+	unsigned int	soft;
+	unsigned int	soft_set;
+	unsigned int	soft_clr;
+	unsigned int	pol;
+	unsigned int	pol_set;
+	unsigned int	pol_clr;
+	unsigned int	dom_en;
+	unsigned int	dbnc_ctrl;
+	unsigned int	dbnc_set;
+	unsigned int	dbnc_clr;
+};
+
+struct mtk_eint_hw {
+	u8		port_mask;
+	u8		ports;
+	unsigned int	ap_num;
+	unsigned int	db_cnt;
+	const unsigned int *db_time;
+};
+
+struct mtk_eint_pin {
+	u16 number;
+	u8 instance;
+	u8 index;
+	bool debounce;
+	bool dual_edge;
+};
+
+extern const unsigned int debounce_time_mt2701[];
+extern const unsigned int debounce_time_mt6765[];
+extern const unsigned int debounce_time_mt6795[];
+extern const unsigned int debounce_time_mt6878[];
+
+struct mtk_eint;
+
+struct mtk_eint_xt {
+	int (*get_gpio_n)(void *data, unsigned long eint_n,
+			  unsigned int *gpio_n,
+			  struct gpio_chip **gpio_chip);
+	int (*get_gpio_state)(void *data, unsigned long eint_n);
+	int (*set_gpio_as_eint)(void *data, unsigned long eint_n);
+};
+
+struct mtk_eint {
+	struct device *dev;
+	void __iomem **base;
+	int nbase;
+	u16 *base_pin_num;
+	struct irq_domain *domain;
+	int irq;
+
+	int *dual_edge;
+	u16 **pin_list;
+	u32 **wake_mask;
+	u32 **cur_mask;
+
+	/* Used to fit into various EINT device */
+	const struct mtk_eint_hw *hw;
+	const struct mtk_eint_regs *regs;
+	struct mtk_eint_pin *pins;
+	u16 num_db_time;
+
+	/* Used to fit into various pinctrl device */
+	void *pctl;
+	const struct mtk_eint_xt *gpio_xlate;
+};
+
+#if IS_ENABLED(CONFIG_EINT_MTK)
+int mtk_eint_do_init(struct mtk_eint *eint, struct mtk_eint_pin *eint_pin);
+int mtk_eint_do_suspend(struct mtk_eint *eint);
+int mtk_eint_do_resume(struct mtk_eint *eint);
+int mtk_eint_set_debounce(struct mtk_eint *eint, unsigned long eint_n,
+			  unsigned int debounce);
+int mtk_eint_find_irq(struct mtk_eint *eint, unsigned long eint_n);
+
+#else
+static inline int mtk_eint_do_init(struct mtk_eint *eint,
+				   struct mtk_eint_pin *eint_pin)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int mtk_eint_do_suspend(struct mtk_eint *eint)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int mtk_eint_do_resume(struct mtk_eint *eint)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int mtk_eint_set_debounce(struct mtk_eint *eint, unsigned long eint_n,
+			  unsigned int debounce)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int mtk_eint_find_irq(struct mtk_eint *eint, unsigned long eint_n)
+{
+	return -EOPNOTSUPP;
+}
+#endif
+#endif /* __MTK_EINT_H */