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, 193 insertions, 0 deletions

diff --git a/drivers/bluetooth/btmrvl_debugfs.c b/drivers/bluetooth/btmrvl_debugfs.c
new file mode 100644
index 000000000..9b3e768c0
--- /dev/null
+++ b/drivers/bluetooth/btmrvl_debugfs.c
@@ -0,0 +1,193 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Marvell Bluetooth driver: debugfs related functions
+ *
+ * Copyright (C) 2009, Marvell International Ltd.
+ **/
+
+#include <linux/debugfs.h>
+#include <linux/slab.h>
+
+#include <net/bluetooth/bluetooth.h>
+#include <net/bluetooth/hci_core.h>
+
+#include "btmrvl_drv.h"
+
+struct btmrvl_debugfs_data {
+	struct dentry *config_dir;
+	struct dentry *status_dir;
+};
+
+static ssize_t btmrvl_hscfgcmd_write(struct file *file,
+			const char __user *ubuf, size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	long result, ret;
+
+	ret = kstrtol_from_user(ubuf, count, 10, &result);
+	if (ret)
+		return ret;
+
+	priv->btmrvl_dev.hscfgcmd = result;
+
+	if (priv->btmrvl_dev.hscfgcmd) {
+		btmrvl_prepare_command(priv);
+		wake_up_interruptible(&priv->main_thread.wait_q);
+	}
+
+	return count;
+}
+
+static ssize_t btmrvl_hscfgcmd_read(struct file *file, char __user *userbuf,
+						size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	char buf[16];
+	int ret;
+
+	ret = snprintf(buf, sizeof(buf) - 1, "%d\n",
+						priv->btmrvl_dev.hscfgcmd);
+
+	return simple_read_from_buffer(userbuf, count, ppos, buf, ret);
+}
+
+static const struct file_operations btmrvl_hscfgcmd_fops = {
+	.read	= btmrvl_hscfgcmd_read,
+	.write	= btmrvl_hscfgcmd_write,
+	.open	= simple_open,
+	.llseek = default_llseek,
+};
+
+static ssize_t btmrvl_pscmd_write(struct file *file, const char __user *ubuf,
+						size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	long result, ret;
+
+	ret = kstrtol_from_user(ubuf, count, 10, &result);
+	if (ret)
+		return ret;
+
+	priv->btmrvl_dev.pscmd = result;
+
+	if (priv->btmrvl_dev.pscmd) {
+		btmrvl_prepare_command(priv);
+		wake_up_interruptible(&priv->main_thread.wait_q);
+	}
+
+	return count;
+
+}
+
+static ssize_t btmrvl_pscmd_read(struct file *file, char __user *userbuf,
+						size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	char buf[16];
+	int ret;
+
+	ret = snprintf(buf, sizeof(buf) - 1, "%d\n", priv->btmrvl_dev.pscmd);
+
+	return simple_read_from_buffer(userbuf, count, ppos, buf, ret);
+}
+
+static const struct file_operations btmrvl_pscmd_fops = {
+	.read = btmrvl_pscmd_read,
+	.write = btmrvl_pscmd_write,
+	.open = simple_open,
+	.llseek = default_llseek,
+};
+
+static ssize_t btmrvl_hscmd_write(struct file *file, const char __user *ubuf,
+						size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	long result, ret;
+
+	ret = kstrtol_from_user(ubuf, count, 10, &result);
+	if (ret)
+		return ret;
+
+	priv->btmrvl_dev.hscmd = result;
+	if (priv->btmrvl_dev.hscmd) {
+		btmrvl_prepare_command(priv);
+		wake_up_interruptible(&priv->main_thread.wait_q);
+	}
+
+	return count;
+}
+
+static ssize_t btmrvl_hscmd_read(struct file *file, char __user *userbuf,
+						size_t count, loff_t *ppos)
+{
+	struct btmrvl_private *priv = file->private_data;
+	char buf[16];
+	int ret;
+
+	ret = snprintf(buf, sizeof(buf) - 1, "%d\n", priv->btmrvl_dev.hscmd);
+
+	return simple_read_from_buffer(userbuf, count, ppos, buf, ret);
+}
+
+static const struct file_operations btmrvl_hscmd_fops = {
+	.read	= btmrvl_hscmd_read,
+	.write	= btmrvl_hscmd_write,
+	.open	= simple_open,
+	.llseek = default_llseek,
+};
+
+void btmrvl_debugfs_init(struct hci_dev *hdev)
+{
+	struct btmrvl_private *priv = hci_get_drvdata(hdev);
+	struct btmrvl_debugfs_data *dbg;
+
+	if (!hdev->debugfs)
+		return;
+
+	dbg = kzalloc_obj(*dbg);
+	priv->debugfs_data = dbg;
+
+	if (!dbg) {
+		BT_ERR("Can not allocate memory for btmrvl_debugfs_data.");
+		return;
+	}
+
+	dbg->config_dir = debugfs_create_dir("config", hdev->debugfs);
+
+	debugfs_create_u8("psmode", 0644, dbg->config_dir,
+			  &priv->btmrvl_dev.psmode);
+	debugfs_create_file("pscmd", 0644, dbg->config_dir,
+			    priv, &btmrvl_pscmd_fops);
+	debugfs_create_x16("gpiogap", 0644, dbg->config_dir,
+			   &priv->btmrvl_dev.gpio_gap);
+	debugfs_create_u8("hsmode", 0644, dbg->config_dir,
+			  &priv->btmrvl_dev.hsmode);
+	debugfs_create_file("hscmd", 0644, dbg->config_dir,
+			    priv, &btmrvl_hscmd_fops);
+	debugfs_create_file("hscfgcmd", 0644, dbg->config_dir,
+			    priv, &btmrvl_hscfgcmd_fops);
+
+	dbg->status_dir = debugfs_create_dir("status", hdev->debugfs);
+	debugfs_create_u8("curpsmode", 0444, dbg->status_dir,
+			  &priv->adapter->psmode);
+	debugfs_create_u8("psstate", 0444, dbg->status_dir,
+			  &priv->adapter->ps_state);
+	debugfs_create_u8("hsstate", 0444, dbg->status_dir,
+			  &priv->adapter->hs_state);
+	debugfs_create_u8("txdnldready", 0444, dbg->status_dir,
+			  &priv->btmrvl_dev.tx_dnld_rdy);
+}
+
+void btmrvl_debugfs_remove(struct hci_dev *hdev)
+{
+	struct btmrvl_private *priv = hci_get_drvdata(hdev);
+	struct btmrvl_debugfs_data *dbg = priv->debugfs_data;
+
+	if (!dbg)
+		return;
+
+	debugfs_remove_recursive(dbg->config_dir);
+	debugfs_remove_recursive(dbg->status_dir);
+
+	kfree(dbg);
+}