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, 108 insertions, 0 deletions

diff --git a/net/openvswitch/conntrack.h b/net/openvswitch/conntrack.h
new file mode 100644
index 000000000..ab21d032f
--- /dev/null
+++ b/net/openvswitch/conntrack.h
@@ -0,0 +1,108 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2015 Nicira, Inc.
+ */
+
+#ifndef OVS_CONNTRACK_H
+#define OVS_CONNTRACK_H 1
+
+#include "flow.h"
+
+struct ovs_conntrack_info;
+struct ovs_ct_limit_info;
+enum ovs_key_attr;
+
+#if IS_ENABLED(CONFIG_NF_CONNTRACK)
+int ovs_ct_init(struct net *);
+void ovs_ct_exit_start(struct net *net);
+void ovs_ct_exit_finish(struct net *net);
+bool ovs_ct_verify(struct net *, enum ovs_key_attr attr);
+int ovs_ct_copy_action(struct net *, const struct nlattr *,
+		       const struct sw_flow_key *, struct sw_flow_actions **,
+		       bool log);
+int ovs_ct_action_to_attr(const struct ovs_conntrack_info *, struct sk_buff *);
+
+int ovs_ct_execute(struct net *, struct sk_buff *, struct sw_flow_key *,
+		   const struct ovs_conntrack_info *);
+int ovs_ct_clear(struct sk_buff *skb, struct sw_flow_key *key);
+
+void ovs_ct_fill_key(const struct sk_buff *skb, struct sw_flow_key *key,
+		     bool post_ct);
+int ovs_ct_put_key(const struct sw_flow_key *swkey,
+		   const struct sw_flow_key *output, struct sk_buff *skb);
+void ovs_ct_free_action(const struct nlattr *a);
+
+#define CT_SUPPORTED_MASK (OVS_CS_F_NEW | OVS_CS_F_ESTABLISHED | \
+			   OVS_CS_F_RELATED | OVS_CS_F_REPLY_DIR | \
+			   OVS_CS_F_INVALID | OVS_CS_F_TRACKED | \
+			   OVS_CS_F_SRC_NAT | OVS_CS_F_DST_NAT)
+#else
+#include <linux/errno.h>
+
+static inline int ovs_ct_init(struct net *net) { return 0; }
+
+static inline void ovs_ct_exit_start(struct net *net) { }
+static inline void ovs_ct_exit_finish(struct net *net) { }
+
+static inline bool ovs_ct_verify(struct net *net, int attr)
+{
+	return false;
+}
+
+static inline int ovs_ct_copy_action(struct net *net, const struct nlattr *nla,
+				     const struct sw_flow_key *key,
+				     struct sw_flow_actions **acts, bool log)
+{
+	return -ENOTSUPP;
+}
+
+static inline int ovs_ct_action_to_attr(const struct ovs_conntrack_info *info,
+					struct sk_buff *skb)
+{
+	return -ENOTSUPP;
+}
+
+static inline int ovs_ct_execute(struct net *net, struct sk_buff *skb,
+				 struct sw_flow_key *key,
+				 const struct ovs_conntrack_info *info)
+{
+	kfree_skb(skb);
+	return -ENOTSUPP;
+}
+
+static inline int ovs_ct_clear(struct sk_buff *skb,
+			       struct sw_flow_key *key)
+{
+	return -ENOTSUPP;
+}
+
+static inline void ovs_ct_fill_key(const struct sk_buff *skb,
+				   struct sw_flow_key *key,
+				   bool post_ct)
+{
+	key->ct_state = 0;
+	key->ct_zone = 0;
+	key->ct.mark = 0;
+	memset(&key->ct.labels, 0, sizeof(key->ct.labels));
+	/* Clear 'ct_orig_proto' to mark the non-existence of original
+	 * direction key fields.
+	 */
+	key->ct_orig_proto = 0;
+}
+
+static inline int ovs_ct_put_key(const struct sw_flow_key *swkey,
+				 const struct sw_flow_key *output,
+				 struct sk_buff *skb)
+{
+	return 0;
+}
+
+static inline void ovs_ct_free_action(const struct nlattr *a) { }
+
+#define CT_SUPPORTED_MASK 0
+#endif /* CONFIG_NF_CONNTRACK */
+
+#if IS_ENABLED(CONFIG_NETFILTER_CONNCOUNT)
+extern struct genl_family dp_ct_limit_genl_family;
+#endif
+#endif /* ovs_conntrack.h */