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, 132 insertions, 0 deletions

diff --git a/drivers/pinctrl/pinctrl-equilibrium.h b/drivers/pinctrl/pinctrl-equilibrium.h
new file mode 100644
index 000000000..b56124d7f
--- /dev/null
+++ b/drivers/pinctrl/pinctrl-equilibrium.h
@@ -0,0 +1,132 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ *  Copyright(c) 2019 Intel Corporation.
+ */
+
+#ifndef __PINCTRL_EQUILIBRIUM_H
+#define __PINCTRL_EQUILIBRIUM_H
+
+/* PINPAD register offset */
+#define REG_PMX_BASE	0x0	/* Port Multiplexer Control Register */
+#define REG_PUEN	0x80	/* PULL UP Enable Register */
+#define REG_PDEN	0x84	/* PULL DOWN Enable Register */
+#define REG_SRC		0x88	/* Slew Rate Control Register */
+#define REG_DCC0	0x8C	/* Drive Current Control Register 0 */
+#define REG_DCC1	0x90	/* Drive Current Control Register 1 */
+#define REG_OD		0x94	/* Open Drain Enable Register */
+#define REG_AVAIL	0x98	/* Pad Control Availability Register */
+#define DRV_CUR_PINS	16	/* Drive Current pin number per register */
+#define REG_DRCC(x)	(REG_DCC0 + (x) * 4) /* Driver current macro */
+
+/* GPIO register offset */
+#define GPIO_OUT	0x0	/* Data Output Register */
+#define GPIO_IN		0x4	/* Data Input Register */
+#define GPIO_DIR	0x8	/* Direction Register */
+#define GPIO_EXINTCR0	0x18	/* External Interrupt Control Register 0 */
+#define GPIO_EXINTCR1	0x1C	/* External Interrupt Control Register 1 */
+#define GPIO_IRNCR	0x20	/* IRN Capture Register */
+#define GPIO_IRNICR	0x24	/* IRN Interrupt Control Register */
+#define GPIO_IRNEN	0x28	/* IRN Interrupt Enable Register */
+#define GPIO_IRNCFG	0x2C	/* IRN Interrupt Configuration Register */
+#define GPIO_IRNRNSET	0x30	/* IRN Interrupt Enable Set Register */
+#define GPIO_IRNENCLR	0x34	/* IRN Interrupt Enable Clear Register */
+#define GPIO_OUTSET	0x40	/* Output Set Register */
+#define GPIO_OUTCLR	0x44	/* Output Clear Register */
+#define GPIO_DIRSET	0x48	/* Direction Set Register */
+#define GPIO_DIRCLR	0x4C	/* Direction Clear Register */
+
+/* parse given pin's driver current value */
+#define PARSE_DRV_CURRENT(val, pin) (((val) >> ((pin) * 2)) & 0x3)
+
+#define GPIO_EDGE_TRIG		0
+#define GPIO_LEVEL_TRIG		1
+#define GPIO_SINGLE_EDGE	0
+#define GPIO_BOTH_EDGE		1
+#define GPIO_POSITIVE_TRIG	0
+#define GPIO_NEGATIVE_TRIG	1
+
+#define EQBR_GPIO_MODE		0
+
+typedef enum {
+	OP_COUNT_NR_FUNCS,
+	OP_ADD_FUNCS,
+	OP_COUNT_NR_FUNC_GRPS,
+	OP_ADD_FUNC_GRPS,
+	OP_NONE,
+} funcs_util_ops;
+
+/**
+ * struct gpio_irq_type: gpio irq configuration
+ * @trig_type: level trigger or edge trigger
+ * @edge_type: sigle edge or both edge
+ * @logic_type: positive trigger or negative trigger
+ */
+struct gpio_irq_type {
+	unsigned int trig_type;
+	unsigned int edge_type;
+	unsigned int logic_type;
+};
+
+/**
+ * struct eqbr_pin_bank: represent a pin bank.
+ * @membase: base address of the pin bank register.
+ * @id: bank id, to idenify the unique bank.
+ * @pin_base: starting pin number of the pin bank.
+ * @nr_pins: number of the pins of the pin bank.
+ * @aval_pinmap: available pin bitmap of the pin bank.
+ */
+struct eqbr_pin_bank {
+	void __iomem		*membase;
+	unsigned int		id;
+	unsigned int		pin_base;
+	unsigned int		nr_pins;
+	u32			aval_pinmap;
+};
+
+struct fwnode_handle;
+
+/**
+ * struct eqbr_gpio_ctrl: represent a gpio controller.
+ * @chip: gpio chip.
+ * @fwnode: firmware node of gpio controller.
+ * @bank: pointer to corresponding pin bank.
+ * @membase: base address of the gpio controller.
+ * @name: gpio chip name.
+ * @virq: irq number of the gpio chip to parent's irq domain.
+ * @lock: spin lock to protect gpio register write.
+ */
+struct eqbr_gpio_ctrl {
+	struct gpio_generic_chip chip;
+	struct fwnode_handle	*fwnode;
+	struct eqbr_pin_bank	*bank;
+	void __iomem		*membase;
+	const char		*name;
+	unsigned int		virq;
+	raw_spinlock_t		lock; /* protect gpio register */
+};
+
+/**
+ * struct eqbr_pinctrl_drv_data:
+ * @dev: device instance representing the controller.
+ * @pctl_desc: pin controller descriptor.
+ * @pctl_dev: pin control class device
+ * @membase: base address of pin controller
+ * @pin_banks: list of pin banks of the driver.
+ * @nr_banks: number of pin banks.
+ * @gpio_ctrls: list of gpio controllers.
+ * @nr_gpio_ctrls: number of gpio controllers.
+ * @lock: protect pinctrl register write
+ */
+struct eqbr_pinctrl_drv_data {
+	struct device			*dev;
+	struct pinctrl_desc		pctl_desc;
+	struct pinctrl_dev		*pctl_dev;
+	void __iomem			*membase;
+	struct eqbr_pin_bank		*pin_banks;
+	unsigned int			nr_banks;
+	struct eqbr_gpio_ctrl		*gpio_ctrls;
+	unsigned int			nr_gpio_ctrls;
+	raw_spinlock_t			lock; /* protect pinpad register */
+};
+
+#endif /* __PINCTRL_EQUILIBRIUM_H */