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, 95 insertions, 0 deletions

diff --git a/net/devres.c b/net/devres.c
new file mode 100644
index 000000000..5ccf6ca31
--- /dev/null
+++ b/net/devres.c
@@ -0,0 +1,95 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * This file contains all networking devres helpers.
+ */
+
+#include <linux/device.h>
+#include <linux/etherdevice.h>
+#include <linux/netdevice.h>
+
+struct net_device_devres {
+	struct net_device *ndev;
+};
+
+static void devm_free_netdev(struct device *dev, void *this)
+{
+	struct net_device_devres *res = this;
+
+	free_netdev(res->ndev);
+}
+
+struct net_device *devm_alloc_etherdev_mqs(struct device *dev, int sizeof_priv,
+					   unsigned int txqs, unsigned int rxqs)
+{
+	struct net_device_devres *dr;
+
+	dr = devres_alloc(devm_free_netdev, sizeof(*dr), GFP_KERNEL);
+	if (!dr)
+		return NULL;
+
+	dr->ndev = alloc_etherdev_mqs(sizeof_priv, txqs, rxqs);
+	if (!dr->ndev) {
+		devres_free(dr);
+		return NULL;
+	}
+
+	devres_add(dev, dr);
+
+	return dr->ndev;
+}
+EXPORT_SYMBOL(devm_alloc_etherdev_mqs);
+
+static void devm_unregister_netdev(struct device *dev, void *this)
+{
+	struct net_device_devres *res = this;
+
+	unregister_netdev(res->ndev);
+}
+
+static int netdev_devres_match(struct device *dev, void *this, void *match_data)
+{
+	struct net_device_devres *res = this;
+	struct net_device *ndev = match_data;
+
+	return ndev == res->ndev;
+}
+
+/**
+ *	devm_register_netdev - resource managed variant of register_netdev()
+ *	@dev: managing device for this netdev - usually the parent device
+ *	@ndev: device to register
+ *
+ *	This is a devres variant of register_netdev() for which the unregister
+ *	function will be called automatically when the managing device is
+ *	detached. Note: the net_device used must also be resource managed by
+ *	the same struct device.
+ */
+int devm_register_netdev(struct device *dev, struct net_device *ndev)
+{
+	struct net_device_devres *dr;
+	int ret;
+
+	/* struct net_device must itself be managed. For now a managed netdev
+	 * can only be allocated by devm_alloc_etherdev_mqs() so the check is
+	 * straightforward.
+	 */
+	if (WARN_ON(!devres_find(dev, devm_free_netdev,
+				 netdev_devres_match, ndev)))
+		return -EINVAL;
+
+	dr = devres_alloc(devm_unregister_netdev, sizeof(*dr), GFP_KERNEL);
+	if (!dr)
+		return -ENOMEM;
+
+	ret = register_netdev(ndev);
+	if (ret) {
+		devres_free(dr);
+		return ret;
+	}
+
+	dr->ndev = ndev;
+	devres_add(ndev->dev.parent, dr);
+
+	return 0;
+}
+EXPORT_SYMBOL(devm_register_netdev);