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, 97 insertions, 0 deletions

diff --git a/include/kunit/test-bug.h b/include/kunit/test-bug.h
new file mode 100644
index 000000000..99869029f
--- /dev/null
+++ b/include/kunit/test-bug.h
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * KUnit API providing hooks for non-test code to interact with tests.
+ *
+ * Copyright (C) 2020, Google LLC.
+ * Author: Uriel Guajardo <urielguajardo@google.com>
+ */
+
+#ifndef _KUNIT_TEST_BUG_H
+#define _KUNIT_TEST_BUG_H
+
+#include <linux/stddef.h> /* for NULL */
+#include <linux/types.h>  /* for bool */
+
+#if IS_ENABLED(CONFIG_KUNIT)
+
+#include <linux/jump_label.h> /* For static branch */
+#include <linux/sched.h>
+
+/* Static key if KUnit is running any tests. */
+DECLARE_STATIC_KEY_FALSE(kunit_running);
+
+/* Hooks table: a table of function pointers filled in when kunit loads */
+extern struct kunit_hooks_table {
+	__printf(3, 4) void (*fail_current_test)(const char*, int, const char*, ...);
+	void *(*get_static_stub_address)(struct kunit *test, void *real_fn_addr);
+	bool (*is_suppressed_warning)(bool count);
+} kunit_hooks;
+
+/**
+ * kunit_get_current_test() - Return a pointer to the currently running
+ *			      KUnit test.
+ *
+ * If a KUnit test is running in the current task, returns a pointer to its
+ * associated struct kunit. This pointer can then be passed to any KUnit
+ * function or assertion. If no test is running (or a test is running in a
+ * different task), returns NULL.
+ *
+ * This function is safe to call even when KUnit is disabled. If CONFIG_KUNIT
+ * is not enabled, it will compile down to nothing and will return quickly no
+ * test is running.
+ */
+static inline struct kunit *kunit_get_current_test(void)
+{
+	if (!static_branch_unlikely(&kunit_running))
+		return NULL;
+
+	return current->kunit_test;
+}
+
+
+/**
+ * kunit_fail_current_test() - If a KUnit test is running, fail it.
+ *
+ * If a KUnit test is running in the current task, mark that test as failed.
+ */
+#define kunit_fail_current_test(fmt, ...) do {					\
+		if (static_branch_unlikely(&kunit_running)) {			\
+			/* Guaranteed to be non-NULL when kunit_running true*/	\
+			kunit_hooks.fail_current_test(__FILE__, __LINE__,	\
+						  fmt, ##__VA_ARGS__);		\
+		}								\
+	} while (0)
+
+/**
+ * kunit_is_suppressed_warning() - Check if warnings are being suppressed
+ *                                 by the current KUnit test.
+ * @count: if true, increment the suppression counter on match.
+ *
+ * Returns true if the current task has active warning suppression.
+ * Uses the kunit_running static branch for zero overhead when no tests run.
+ *
+ * A single WARN*() may traverse multiple call sites in the warning path
+ * (e.g., __warn_printk() and __report_bug()). Pass @count = true at the
+ * primary suppression point to count each warning exactly once, and
+ * @count = false at secondary points to suppress output without
+ * inflating the count.
+ */
+static inline bool kunit_is_suppressed_warning(bool count)
+{
+	if (!static_branch_unlikely(&kunit_running))
+		return false;
+
+	return kunit_hooks.is_suppressed_warning &&
+	       kunit_hooks.is_suppressed_warning(count);
+}
+
+#else
+
+static inline struct kunit *kunit_get_current_test(void) { return NULL; }
+static inline bool kunit_is_suppressed_warning(bool count) { return false; }
+
+#define kunit_fail_current_test(fmt, ...) do {} while (0)
+
+#endif
+
+#endif /* _KUNIT_TEST_BUG_H */