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, 226 insertions, 0 deletions

diff --git a/drivers/clk/ux500/clk-sysctrl.c b/drivers/clk/ux500/clk-sysctrl.c
new file mode 100644
index 000000000..ba3258c88
--- /dev/null
+++ b/drivers/clk/ux500/clk-sysctrl.c
@@ -0,0 +1,226 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Sysctrl clock implementation for ux500 platform.
+ *
+ * Copyright (C) 2013 ST-Ericsson SA
+ * Author: Ulf Hansson <ulf.hansson@linaro.org>
+ */
+
+#include <linux/clk-provider.h>
+#include <linux/mfd/abx500/ab8500-sysctrl.h>
+#include <linux/device.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+#include <linux/io.h>
+#include <linux/err.h>
+#include "clk.h"
+
+#define SYSCTRL_MAX_NUM_PARENTS 4
+
+#define to_clk_sysctrl(_hw) container_of(_hw, struct clk_sysctrl, hw)
+
+struct clk_sysctrl {
+	struct clk_hw hw;
+	struct device *dev;
+	u8 parent_index;
+	u16 reg_sel[SYSCTRL_MAX_NUM_PARENTS];
+	u8 reg_mask[SYSCTRL_MAX_NUM_PARENTS];
+	u8 reg_bits[SYSCTRL_MAX_NUM_PARENTS];
+	unsigned long rate;
+	unsigned long enable_delay_us;
+};
+
+/* Sysctrl clock operations. */
+
+static int clk_sysctrl_prepare(struct clk_hw *hw)
+{
+	int ret;
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+
+	ret = ab8500_sysctrl_write(clk->reg_sel[0], clk->reg_mask[0],
+				clk->reg_bits[0]);
+
+	if (!ret && clk->enable_delay_us)
+		usleep_range(clk->enable_delay_us, clk->enable_delay_us +
+			     (clk->enable_delay_us >> 2));
+
+	return ret;
+}
+
+static void clk_sysctrl_unprepare(struct clk_hw *hw)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	if (ab8500_sysctrl_clear(clk->reg_sel[0], clk->reg_mask[0]))
+		dev_err(clk->dev, "clk_sysctrl: %s fail to clear %s.\n",
+			__func__, clk_hw_get_name(hw));
+}
+
+static unsigned long clk_sysctrl_recalc_rate(struct clk_hw *hw,
+					unsigned long parent_rate)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	return clk->rate;
+}
+
+static int clk_sysctrl_set_parent(struct clk_hw *hw, u8 index)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	u8 old_index = clk->parent_index;
+	int ret = 0;
+
+	if (clk->reg_sel[old_index]) {
+		ret = ab8500_sysctrl_clear(clk->reg_sel[old_index],
+					clk->reg_mask[old_index]);
+		if (ret)
+			return ret;
+	}
+
+	if (clk->reg_sel[index]) {
+		ret = ab8500_sysctrl_write(clk->reg_sel[index],
+					clk->reg_mask[index],
+					clk->reg_bits[index]);
+		if (ret) {
+			if (clk->reg_sel[old_index])
+				ab8500_sysctrl_write(clk->reg_sel[old_index],
+						clk->reg_mask[old_index],
+						clk->reg_bits[old_index]);
+			return ret;
+		}
+	}
+	clk->parent_index = index;
+
+	return ret;
+}
+
+static u8 clk_sysctrl_get_parent(struct clk_hw *hw)
+{
+	struct clk_sysctrl *clk = to_clk_sysctrl(hw);
+	return clk->parent_index;
+}
+
+static const struct clk_ops clk_sysctrl_gate_ops = {
+	.prepare = clk_sysctrl_prepare,
+	.unprepare = clk_sysctrl_unprepare,
+};
+
+static const struct clk_ops clk_sysctrl_gate_fixed_rate_ops = {
+	.prepare = clk_sysctrl_prepare,
+	.unprepare = clk_sysctrl_unprepare,
+	.recalc_rate = clk_sysctrl_recalc_rate,
+};
+
+static const struct clk_ops clk_sysctrl_set_parent_ops = {
+	.determine_rate = clk_hw_determine_rate_no_reparent,
+	.set_parent = clk_sysctrl_set_parent,
+	.get_parent = clk_sysctrl_get_parent,
+};
+
+static struct clk *clk_reg_sysctrl(struct device *dev,
+				const char *name,
+				const char **parent_names,
+				u8 num_parents,
+				u16 *reg_sel,
+				u8 *reg_mask,
+				u8 *reg_bits,
+				unsigned long rate,
+				unsigned long enable_delay_us,
+				unsigned long flags,
+				const struct clk_ops *clk_sysctrl_ops)
+{
+	struct clk_sysctrl *clk;
+	struct clk_init_data clk_sysctrl_init;
+	struct clk *clk_reg;
+	int i;
+
+	if (!dev)
+		return ERR_PTR(-EINVAL);
+
+	if (!name || (num_parents > SYSCTRL_MAX_NUM_PARENTS)) {
+		dev_err(dev, "clk_sysctrl: invalid arguments passed\n");
+		return ERR_PTR(-EINVAL);
+	}
+
+	clk = devm_kzalloc(dev, sizeof(*clk), GFP_KERNEL);
+	if (!clk)
+		return ERR_PTR(-ENOMEM);
+
+	/* set main clock registers */
+	clk->reg_sel[0] = reg_sel[0];
+	clk->reg_bits[0] = reg_bits[0];
+	clk->reg_mask[0] = reg_mask[0];
+
+	/* handle clocks with more than one parent */
+	for (i = 1; i < num_parents; i++) {
+		clk->reg_sel[i] = reg_sel[i];
+		clk->reg_bits[i] = reg_bits[i];
+		clk->reg_mask[i] = reg_mask[i];
+	}
+
+	clk->parent_index = 0;
+	clk->rate = rate;
+	clk->enable_delay_us = enable_delay_us;
+	clk->dev = dev;
+
+	clk_sysctrl_init.name = name;
+	clk_sysctrl_init.ops = clk_sysctrl_ops;
+	clk_sysctrl_init.flags = flags;
+	clk_sysctrl_init.parent_names = parent_names;
+	clk_sysctrl_init.num_parents = num_parents;
+	clk->hw.init = &clk_sysctrl_init;
+
+	clk_reg = devm_clk_register(clk->dev, &clk->hw);
+	if (IS_ERR(clk_reg))
+		dev_err(dev, "clk_sysctrl: clk_register failed\n");
+
+	return clk_reg;
+}
+
+struct clk *clk_reg_sysctrl_gate(struct device *dev,
+				const char *name,
+				const char *parent_name,
+				u16 reg_sel,
+				u8 reg_mask,
+				u8 reg_bits,
+				unsigned long enable_delay_us,
+				unsigned long flags)
+{
+	const char **parent_names = (parent_name ? &parent_name : NULL);
+	u8 num_parents = (parent_name ? 1 : 0);
+
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			&reg_sel, &reg_mask, &reg_bits, 0, enable_delay_us,
+			flags, &clk_sysctrl_gate_ops);
+}
+
+struct clk *clk_reg_sysctrl_gate_fixed_rate(struct device *dev,
+					const char *name,
+					const char *parent_name,
+					u16 reg_sel,
+					u8 reg_mask,
+					u8 reg_bits,
+					unsigned long rate,
+					unsigned long enable_delay_us,
+					unsigned long flags)
+{
+	const char **parent_names = (parent_name ? &parent_name : NULL);
+	u8 num_parents = (parent_name ? 1 : 0);
+
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			&reg_sel, &reg_mask, &reg_bits,
+			rate, enable_delay_us, flags,
+			&clk_sysctrl_gate_fixed_rate_ops);
+}
+
+struct clk *clk_reg_sysctrl_set_parent(struct device *dev,
+				const char *name,
+				const char **parent_names,
+				u8 num_parents,
+				u16 *reg_sel,
+				u8 *reg_mask,
+				u8 *reg_bits,
+				unsigned long flags)
+{
+	return clk_reg_sysctrl(dev, name, parent_names, num_parents,
+			reg_sel, reg_mask, reg_bits, 0, 0, flags,
+			&clk_sysctrl_set_parent_ops);
+}