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, 159 insertions, 0 deletions

diff --git a/include/net/net_debug.h b/include/net/net_debug.h
new file mode 100644
index 000000000..47f7a4a87
--- /dev/null
+++ b/include/net/net_debug.h
@@ -0,0 +1,159 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _LINUX_NET_DEBUG_H
+#define _LINUX_NET_DEBUG_H
+
+#include <linux/bug.h>
+#include <linux/kern_levels.h>
+
+struct net_device;
+
+__printf(3, 4) __cold
+void netdev_printk(const char *level, const struct net_device *dev,
+		   const char *format, ...);
+__printf(2, 3) __cold
+void netdev_emerg(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_alert(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_crit(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_err(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_warn(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_notice(const struct net_device *dev, const char *format, ...);
+__printf(2, 3) __cold
+void netdev_info(const struct net_device *dev, const char *format, ...);
+
+#define netdev_level_once(level, dev, fmt, ...)			\
+do {								\
+	static bool __section(".data..once") __print_once;	\
+								\
+	if (!__print_once) {					\
+		__print_once = true;				\
+		netdev_printk(level, dev, fmt, ##__VA_ARGS__);	\
+	}							\
+} while (0)
+
+#define netdev_emerg_once(dev, fmt, ...) \
+	netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
+#define netdev_alert_once(dev, fmt, ...) \
+	netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
+#define netdev_crit_once(dev, fmt, ...) \
+	netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
+#define netdev_err_once(dev, fmt, ...) \
+	netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
+#define netdev_warn_once(dev, fmt, ...) \
+	netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
+#define netdev_notice_once(dev, fmt, ...) \
+	netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
+#define netdev_info_once(dev, fmt, ...) \
+	netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
+
+#if defined(CONFIG_DYNAMIC_DEBUG) || \
+	(defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
+#define netdev_dbg(__dev, format, args...)			\
+do {								\
+	dynamic_netdev_dbg(__dev, format, ##args);		\
+} while (0)
+#elif defined(DEBUG)
+#define netdev_dbg(__dev, format, args...)			\
+	netdev_printk(KERN_DEBUG, __dev, format, ##args)
+#else
+#define netdev_dbg(__dev, format, args...)			\
+({								\
+	if (0)							\
+		netdev_printk(KERN_DEBUG, __dev, format, ##args); \
+})
+#endif
+
+#if defined(VERBOSE_DEBUG)
+#define netdev_vdbg	netdev_dbg
+#else
+
+#define netdev_vdbg(dev, format, args...)			\
+({								\
+	if (0)							\
+		netdev_printk(KERN_DEBUG, dev, format, ##args);	\
+	0;							\
+})
+#endif
+
+/* netif printk helpers, similar to netdev_printk */
+
+#define netif_printk(priv, type, level, dev, fmt, args...)	\
+do {					  			\
+	if (netif_msg_##type(priv))				\
+		netdev_printk(level, (dev), fmt, ##args);	\
+} while (0)
+
+#define netif_level(level, priv, type, dev, fmt, args...)	\
+do {								\
+	if (netif_msg_##type(priv))				\
+		netdev_##level(dev, fmt, ##args);		\
+} while (0)
+
+#define netif_emerg(priv, type, dev, fmt, args...)		\
+	netif_level(emerg, priv, type, dev, fmt, ##args)
+#define netif_alert(priv, type, dev, fmt, args...)		\
+	netif_level(alert, priv, type, dev, fmt, ##args)
+#define netif_crit(priv, type, dev, fmt, args...)		\
+	netif_level(crit, priv, type, dev, fmt, ##args)
+#define netif_err(priv, type, dev, fmt, args...)		\
+	netif_level(err, priv, type, dev, fmt, ##args)
+#define netif_warn(priv, type, dev, fmt, args...)		\
+	netif_level(warn, priv, type, dev, fmt, ##args)
+#define netif_notice(priv, type, dev, fmt, args...)		\
+	netif_level(notice, priv, type, dev, fmt, ##args)
+#define netif_info(priv, type, dev, fmt, args...)		\
+	netif_level(info, priv, type, dev, fmt, ##args)
+
+#if defined(CONFIG_DYNAMIC_DEBUG) || \
+	(defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
+#define netif_dbg(priv, type, netdev, format, args...)		\
+do {								\
+	if (netif_msg_##type(priv))				\
+		dynamic_netdev_dbg(netdev, format, ##args);	\
+} while (0)
+#elif defined(DEBUG)
+#define netif_dbg(priv, type, dev, format, args...)		\
+	netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
+#else
+#define netif_dbg(priv, type, dev, format, args...)			\
+({									\
+	if (0)								\
+		netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
+	0;								\
+})
+#endif
+
+/* if @cond then downgrade to debug, else print at @level */
+#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...)     \
+	do {                                                              \
+		if (cond)                                                 \
+			netif_dbg(priv, type, netdev, fmt, ##args);       \
+		else                                                      \
+			netif_ ## level(priv, type, netdev, fmt, ##args); \
+	} while (0)
+
+#if defined(VERBOSE_DEBUG)
+#define netif_vdbg	netif_dbg
+#else
+#define netif_vdbg(priv, type, dev, format, args...)		\
+({								\
+	if (0)							\
+		netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
+	0;							\
+})
+#endif
+
+
+#if defined(CONFIG_DEBUG_NET)
+#define DEBUG_NET_WARN_ON_ONCE(cond) ((void)WARN_ON_ONCE(cond))
+#define DEBUG_NET_WARN_ONCE(cond, format...) ((void)WARN_ONCE(cond, format))
+#else
+#define DEBUG_NET_WARN_ON_ONCE(cond) BUILD_BUG_ON_INVALID(cond)
+#define DEBUG_NET_WARN_ONCE(cond, format...) BUILD_BUG_ON_INVALID(cond)
+#endif
+
+#endif	/* _LINUX_NET_DEBUG_H */