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, 59 insertions, 0 deletions

diff --git a/drivers/clk/qcom/reset.c b/drivers/clk/qcom/reset.c
new file mode 100644
index 000000000..d96c96a90
--- /dev/null
+++ b/drivers/clk/qcom/reset.c
@@ -0,0 +1,59 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2013, The Linux Foundation. All rights reserved.
+ */
+
+#include <linux/bitops.h>
+#include <linux/export.h>
+#include <linux/regmap.h>
+#include <linux/reset-controller.h>
+#include <linux/delay.h>
+
+#include "reset.h"
+
+static int qcom_reset(struct reset_controller_dev *rcdev, unsigned long id)
+{
+	struct qcom_reset_controller *rst = to_qcom_reset_controller(rcdev);
+
+	rcdev->ops->assert(rcdev, id);
+	fsleep(rst->reset_map[id].udelay ?: 1); /* use 1 us as default */
+
+	rcdev->ops->deassert(rcdev, id);
+	return 0;
+}
+
+static int qcom_reset_set_assert(struct reset_controller_dev *rcdev,
+				 unsigned long id, bool assert)
+{
+	struct qcom_reset_controller *rst;
+	const struct qcom_reset_map *map;
+	u32 mask;
+
+	rst = to_qcom_reset_controller(rcdev);
+	map = &rst->reset_map[id];
+	mask = map->bitmask ? map->bitmask : BIT(map->bit);
+
+	regmap_update_bits(rst->regmap, map->reg, mask, assert ? mask : 0);
+
+	/* Read back the register to ensure write completion, ignore the value */
+	regmap_read(rst->regmap, map->reg, &mask);
+
+	return 0;
+}
+
+static int qcom_reset_assert(struct reset_controller_dev *rcdev, unsigned long id)
+{
+	return qcom_reset_set_assert(rcdev, id, true);
+}
+
+static int qcom_reset_deassert(struct reset_controller_dev *rcdev, unsigned long id)
+{
+	return qcom_reset_set_assert(rcdev, id, false);
+}
+
+const struct reset_control_ops qcom_reset_ops = {
+	.reset = qcom_reset,
+	.assert = qcom_reset_assert,
+	.deassert = qcom_reset_deassert,
+};
+EXPORT_SYMBOL_GPL(qcom_reset_ops);