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, 65 insertions, 0 deletions

diff --git a/arch/powerpc/lib/locks.c b/arch/powerpc/lib/locks.c
new file mode 100644
index 000000000..04165b7a1
--- /dev/null
+++ b/arch/powerpc/lib/locks.c
@@ -0,0 +1,65 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Spin and read/write lock operations.
+ *
+ * Copyright (C) 2001-2004 Paul Mackerras <paulus@au.ibm.com>, IBM
+ * Copyright (C) 2001 Anton Blanchard <anton@au.ibm.com>, IBM
+ * Copyright (C) 2002 Dave Engebretsen <engebret@us.ibm.com>, IBM
+ *   Rework to support virtual processors
+ */
+
+#include <linux/kernel.h>
+#include <linux/spinlock.h>
+#include <linux/export.h>
+#include <linux/smp.h>
+
+/* waiting for a spinlock... */
+#if defined(CONFIG_PPC_SPLPAR)
+#include <asm/hvcall.h>
+#include <asm/smp.h>
+
+void splpar_spin_yield(arch_spinlock_t *lock)
+{
+	unsigned int lock_value, holder_cpu, yield_count;
+
+	lock_value = lock->slock;
+	if (lock_value == 0)
+		return;
+	holder_cpu = lock_value & 0xffff;
+	BUG_ON(holder_cpu >= NR_CPUS);
+
+	yield_count = yield_count_of(holder_cpu);
+	if ((yield_count & 1) == 0)
+		return;		/* virtual cpu is currently running */
+	rmb();
+	if (lock->slock != lock_value)
+		return;		/* something has changed */
+	yield_to_preempted(holder_cpu, yield_count);
+}
+EXPORT_SYMBOL_GPL(splpar_spin_yield);
+
+/*
+ * Waiting for a read lock or a write lock on a rwlock...
+ * This turns out to be the same for read and write locks, since
+ * we only know the holder if it is write-locked.
+ */
+void splpar_rw_yield(arch_rwlock_t *rw)
+{
+	int lock_value;
+	unsigned int holder_cpu, yield_count;
+
+	lock_value = rw->lock;
+	if (lock_value >= 0)
+		return;		/* no write lock at present */
+	holder_cpu = lock_value & 0xffff;
+	BUG_ON(holder_cpu >= NR_CPUS);
+
+	yield_count = yield_count_of(holder_cpu);
+	if ((yield_count & 1) == 0)
+		return;		/* virtual cpu is currently running */
+	rmb();
+	if (rw->lock != lock_value)
+		return;		/* something has changed */
+	yield_to_preempted(holder_cpu, yield_count);
+}
+#endif