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, 121 insertions, 0 deletions

diff --git a/include/kunit/of.h b/include/kunit/of.h
new file mode 100644
index 000000000..75a760a4e
--- /dev/null
+++ b/include/kunit/of.h
@@ -0,0 +1,121 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _KUNIT_OF_H
+#define _KUNIT_OF_H
+
+#include <kunit/test.h>
+
+struct device_node;
+
+#ifdef CONFIG_OF
+
+void of_node_put_kunit(struct kunit *test, struct device_node *node);
+
+#else
+
+static inline
+void of_node_put_kunit(struct kunit *test, struct device_node *node)
+{
+	kunit_skip(test, "requires CONFIG_OF");
+}
+
+#endif /* !CONFIG_OF */
+
+#if defined(CONFIG_OF) && defined(CONFIG_OF_OVERLAY) && defined(CONFIG_OF_EARLY_FLATTREE)
+
+int of_overlay_fdt_apply_kunit(struct kunit *test, void *overlay_fdt,
+			       u32 overlay_fdt_size, int *ovcs_id);
+#else
+
+static inline int
+of_overlay_fdt_apply_kunit(struct kunit *test, void *overlay_fdt,
+			   u32 overlay_fdt_size, int *ovcs_id)
+{
+	kunit_skip(test, "requires CONFIG_OF and CONFIG_OF_OVERLAY and CONFIG_OF_EARLY_FLATTREE for root node");
+	return -EINVAL;
+}
+
+#endif
+
+/**
+ * __of_overlay_apply_kunit() - Test managed of_overlay_fdt_apply() variant
+ * @test: test context
+ * @overlay_begin: start address of overlay to apply
+ * @overlay_end: end address of overlay to apply
+ *
+ * This is mostly internal API. See of_overlay_apply_kunit() for the wrapper
+ * that makes this easier to use.
+ *
+ * Similar to of_overlay_fdt_apply(), except the overlay is managed by the test
+ * case and is automatically removed with of_overlay_remove() after the test
+ * case concludes.
+ *
+ * Return: 0 on success, negative errno on failure
+ */
+static inline int __of_overlay_apply_kunit(struct kunit *test,
+					   u8 *overlay_begin,
+					   const u8 *overlay_end)
+{
+	int unused;
+
+	return of_overlay_fdt_apply_kunit(test, overlay_begin,
+					  overlay_end - overlay_begin,
+					  &unused);
+}
+
+#define of_overlay_begin(overlay_name) __dtbo_##overlay_name##_begin
+#define of_overlay_end(overlay_name) __dtbo_##overlay_name##_end
+
+#define OF_OVERLAY_DECLARE(overlay_name)			\
+	extern uint8_t of_overlay_begin(overlay_name)[];	\
+	extern uint8_t of_overlay_end(overlay_name)[]		\
+
+/**
+ * of_overlay_apply_kunit() - Test managed of_overlay_fdt_apply() for built-in overlays
+ * @test: test context
+ * @overlay_name: name of overlay to apply
+ *
+ * This macro is used to apply a device tree overlay built with the
+ * cmd_dt_S_dtbo rule in scripts/Makefile.lib that has been compiled into the
+ * kernel image or KUnit test module. The overlay is automatically removed when
+ * the test is finished.
+ *
+ * Unit tests that need device tree nodes should compile an overlay file with
+ * @overlay_name\.dtbo.o in their Makefile along with their unit test and then
+ * load the overlay during their test. The @overlay_name matches the filename
+ * of the overlay without the dtbo filename extension. If CONFIG_OF_OVERLAY is
+ * not enabled, the @test will be skipped.
+ *
+ * In the Makefile
+ *
+ * .. code-block:: none
+ *
+ *	obj-$(CONFIG_OF_OVERLAY_KUNIT_TEST) += overlay_test.o kunit_overlay_test.dtbo.o
+ *
+ * In the test
+ *
+ * .. code-block:: c
+ *
+ *	static void of_overlay_kunit_of_overlay_apply(struct kunit *test)
+ *	{
+ *		struct device_node *np;
+ *
+ *		KUNIT_ASSERT_EQ(test, 0,
+ *				of_overlay_apply_kunit(test, kunit_overlay_test));
+ *
+ *		np = of_find_node_by_name(NULL, "test-kunit");
+ *		KUNIT_EXPECT_NOT_ERR_OR_NULL(test, np);
+ *		of_node_put(np);
+ *	}
+ *
+ * Return: 0 on success, negative errno on failure.
+ */
+#define of_overlay_apply_kunit(test, overlay_name)		\
+({								\
+	OF_OVERLAY_DECLARE(overlay_name);			\
+								\
+	__of_overlay_apply_kunit((test),			\
+			of_overlay_begin(overlay_name),		\
+			of_overlay_end(overlay_name));		\
+})
+
+#endif