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, 214 insertions, 0 deletions

diff --git a/drivers/pinctrl/pinconf.h b/drivers/pinctrl/pinconf.h
new file mode 100644
index 000000000..fc2d6cce9
--- /dev/null
+++ b/drivers/pinctrl/pinconf.h
@@ -0,0 +1,214 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Internal interface between the core pin control system and the
+ * pin config portions
+ *
+ * Copyright (C) 2011 ST-Ericsson SA
+ * Written on behalf of Linaro for ST-Ericsson
+ * Based on bits of regulator core, gpio core and clk core
+ *
+ * Author: Linus Walleij <linus.walleij@linaro.org>
+ */
+
+#include <linux/errno.h>
+
+struct dentry;
+struct device_node;
+struct seq_file;
+
+struct pinctrl_dev;
+struct pinctrl_map;
+struct pinctrl_setting;
+
+#ifdef CONFIG_PINCONF
+
+int pinconf_check_ops(struct pinctrl_dev *pctldev);
+int pinconf_validate_map(const struct pinctrl_map *map, int i);
+int pinconf_map_to_setting(const struct pinctrl_map *map,
+			  struct pinctrl_setting *setting);
+void pinconf_free_setting(const struct pinctrl_setting *setting);
+int pinconf_apply_setting(const struct pinctrl_setting *setting);
+
+int pinconf_set_config(struct pinctrl_dev *pctldev, unsigned int pin,
+		       unsigned long *configs, size_t nconfigs);
+
+/*
+ * You will only be interested in these if you're using PINCONF
+ * so don't supply any stubs for these.
+ */
+int pin_config_get_for_pin(struct pinctrl_dev *pctldev, unsigned int pin,
+			   unsigned long *config);
+int pin_config_group_get(const char *dev_name, const char *pin_group,
+			 unsigned long *config);
+
+#else
+
+static inline int pinconf_check_ops(struct pinctrl_dev *pctldev)
+{
+	return 0;
+}
+
+static inline int pinconf_validate_map(const struct pinctrl_map *map, int i)
+{
+	return 0;
+}
+
+static inline int pinconf_map_to_setting(const struct pinctrl_map *map,
+			  struct pinctrl_setting *setting)
+{
+	return 0;
+}
+
+static inline void pinconf_free_setting(const struct pinctrl_setting *setting)
+{
+}
+
+static inline int pinconf_apply_setting(const struct pinctrl_setting *setting)
+{
+	return 0;
+}
+
+static inline int pinconf_set_config(struct pinctrl_dev *pctldev, unsigned int pin,
+				     unsigned long *configs, size_t nconfigs)
+{
+	return -ENOTSUPP;
+}
+
+static inline int pin_config_get_for_pin(struct pinctrl_dev *pctldev, unsigned int pin,
+			   unsigned long *config)
+{
+	return -ENOTSUPP;
+}
+
+#endif
+
+#if defined(CONFIG_PINCONF) && defined(CONFIG_DEBUG_FS)
+
+void pinconf_show_map(struct seq_file *s, const struct pinctrl_map *map);
+void pinconf_show_setting(struct seq_file *s,
+			  const struct pinctrl_setting *setting);
+void pinconf_init_device_debugfs(struct dentry *devroot,
+				 struct pinctrl_dev *pctldev);
+
+#else
+
+static inline void pinconf_show_map(struct seq_file *s,
+				    const struct pinctrl_map *map)
+{
+}
+
+static inline void pinconf_show_setting(struct seq_file *s,
+					const struct pinctrl_setting *setting)
+{
+}
+
+static inline void pinconf_init_device_debugfs(struct dentry *devroot,
+					       struct pinctrl_dev *pctldev)
+{
+}
+
+#endif
+
+/*
+ * The following functions are available if the driver uses the generic
+ * pin config.
+ */
+
+#if defined(CONFIG_GENERIC_PINCONF) && defined(CONFIG_DEBUG_FS)
+
+void pinconf_generic_dump_pins(struct pinctrl_dev *pctldev,
+			       struct seq_file *s, const char *gname,
+			       unsigned int pin);
+
+void pinconf_generic_dump_config(struct pinctrl_dev *pctldev,
+				 struct seq_file *s, unsigned long config);
+#else
+
+static inline void pinconf_generic_dump_pins(struct pinctrl_dev *pctldev,
+					     struct seq_file *s,
+					     const char *gname, unsigned int pin)
+{
+	return;
+}
+
+static inline void pinconf_generic_dump_config(struct pinctrl_dev *pctldev,
+					       struct seq_file *s,
+					       unsigned long config)
+{
+	return;
+}
+#endif
+
+#if defined(CONFIG_GENERIC_PINCONF) && defined(CONFIG_OF)
+int pinconf_generic_parse_dt_config(struct device_node *np,
+				    struct pinctrl_dev *pctldev,
+				    unsigned long **configs,
+				    unsigned int *nconfigs);
+
+int pinconf_generic_parse_dt_pinmux(struct device_node *np, struct device *dev,
+				    unsigned int **pid, unsigned int **pmux,
+				    unsigned int *npins);
+#else
+static inline int
+pinconf_generic_parse_dt_config(struct device_node *np,
+				struct pinctrl_dev *pctldev,
+				unsigned long **configs,
+				unsigned int *nconfigs)
+{
+	return -ENOTSUPP;
+}
+
+static inline int
+pinconf_generic_parse_dt_pinmux(struct device_node *np, struct device *dev,
+				unsigned int **pid, unsigned int **pmux,
+				unsigned int *npins)
+{
+	return -ENOTSUPP;
+}
+#endif
+
+#if defined(CONFIG_GENERIC_PINCTRL)
+int pinctrl_generic_pins_function_dt_node_to_map(struct pinctrl_dev *pctldev,
+						 struct device_node *np,
+						 struct pinctrl_map **maps,
+						 unsigned int *num_maps);
+
+int pinctrl_generic_pinmux_dt_node_to_map(struct pinctrl_dev *pctldev,
+					  struct device_node *np,
+					  struct pinctrl_map **maps,
+					  unsigned int *num_maps);
+
+int pinctrl_generic_to_map(struct pinctrl_dev *pctldev, struct device_node *parent,
+			   struct device_node *np, struct pinctrl_map **maps,
+			   unsigned int *num_maps, unsigned int *num_reserved_maps,
+			   const char **group_name, unsigned int ngroups,
+			   void *data, unsigned int *pins, unsigned int npins);
+#else
+static inline int
+pinctrl_generic_pins_function_dt_node_to_map(struct pinctrl_dev *pctldev,
+					     struct device_node *np,
+					     struct pinctrl_map **maps,
+					     unsigned int *num_maps)
+{
+	return -ENOTSUPP;
+}
+
+static inline int
+pinctrl_generic_pinmux_dt_node_to_map(struct pinctrl_dev *pctldev,
+				      struct device_node *np,
+				      struct pinctrl_map **maps,
+				      unsigned int *num_maps)
+{
+	return -ENOTSUPP;
+}
+
+static inline int
+pinctrl_generic_to_map(struct pinctrl_dev *pctldev, struct device_node *parent,
+		       struct device_node *np, struct pinctrl_map **maps,
+		       unsigned int *num_maps, unsigned int *num_reserved_maps,
+		       const char **group_name, unsigned int ngroups,
+		       void *data, unsigned int *pins, unsigned int npins)
+{
+	return -ENOTSUPP;
+}
+#endif