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, 74 insertions, 0 deletions

diff --git a/net/llc/llc_output.c b/net/llc/llc_output.c
new file mode 100644
index 000000000..5a6466fc6
--- /dev/null
+++ b/net/llc/llc_output.c
@@ -0,0 +1,74 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * llc_output.c - LLC minimal output path
+ *
+ * Copyright (c) 1997 by Procom Technology, Inc.
+ * 		 2001-2003 by Arnaldo Carvalho de Melo <acme@conectiva.com.br>
+ */
+
+#include <linux/if_arp.h>
+#include <linux/netdevice.h>
+#include <linux/skbuff.h>
+#include <linux/export.h>
+#include <net/llc.h>
+#include <net/llc_pdu.h>
+
+/**
+ *	llc_mac_hdr_init - fills MAC header fields
+ *	@skb: Address of the frame to initialize its MAC header
+ *	@sa: The MAC source address
+ *	@da: The MAC destination address
+ *
+ *	Fills MAC header fields, depending on MAC type. Returns 0, If MAC type
+ *	is a valid type and initialization completes correctly 1, otherwise.
+ */
+int llc_mac_hdr_init(struct sk_buff *skb,
+		     const unsigned char *sa, const unsigned char *da)
+{
+	int rc = -EINVAL;
+
+	switch (skb->dev->type) {
+	case ARPHRD_ETHER:
+	case ARPHRD_LOOPBACK:
+		rc = dev_hard_header(skb, skb->dev, ETH_P_802_2, da, sa,
+				     skb->len);
+		if (rc > 0)
+			rc = 0;
+		break;
+	default:
+		break;
+	}
+	return rc;
+}
+
+/**
+ *	llc_build_and_send_ui_pkt - unitdata request interface for upper layers
+ *	@sap: sap to use
+ *	@skb: packet to send
+ *	@dmac: destination mac address
+ *	@dsap: destination sap
+ *
+ *	Upper layers calls this function when upper layer wants to send data
+ *	using connection-less mode communication (UI pdu).
+ *
+ *	Accept data frame from network layer to be sent using connection-
+ *	less mode communication; timeout/retries handled by network layer;
+ *	package primitive as an event and send to SAP event handler
+ */
+int llc_build_and_send_ui_pkt(struct llc_sap *sap, struct sk_buff *skb,
+			      const unsigned char *dmac, unsigned char dsap)
+{
+	int rc;
+	llc_pdu_header_init(skb, LLC_PDU_TYPE_U, sap->laddr.lsap,
+			    dsap, LLC_PDU_CMD);
+	llc_pdu_init_as_ui_cmd(skb);
+	rc = llc_mac_hdr_init(skb, skb->dev->dev_addr, dmac);
+	if (likely(!rc))
+		rc = dev_queue_xmit(skb);
+	else
+		kfree_skb(skb);
+	return rc;
+}
+
+EXPORT_SYMBOL(llc_mac_hdr_init);
+EXPORT_SYMBOL(llc_build_and_send_ui_pkt);