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, 123 insertions, 0 deletions

diff --git a/include/asm-generic/futex.h b/include/asm-generic/futex.h
new file mode 100644
index 000000000..fbbcfd801
--- /dev/null
+++ b/include/asm-generic/futex.h
@@ -0,0 +1,123 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_GENERIC_FUTEX_H
+#define _ASM_GENERIC_FUTEX_H
+
+#include <linux/futex.h>
+#include <linux/uaccess.h>
+#include <asm/errno.h>
+
+#ifndef futex_atomic_cmpxchg_inatomic
+#ifndef CONFIG_SMP
+/*
+ * The following implementation only for uniprocessor machines.
+ * It relies on preempt_disable() ensuring mutual exclusion.
+ *
+ */
+#define futex_atomic_cmpxchg_inatomic(uval, uaddr, oldval, newval) \
+	futex_atomic_cmpxchg_inatomic_local(uval, uaddr, oldval, newval)
+#define arch_futex_atomic_op_inuser(op, oparg, oval, uaddr) \
+	futex_atomic_op_inuser_local(op, oparg, oval, uaddr)
+#endif /* CONFIG_SMP */
+#endif
+
+/**
+ * futex_atomic_op_inuser_local() - Atomic arithmetic operation with constant
+ *			  argument and comparison of the previous
+ *			  futex value with another constant.
+ *
+ * @op:		operation to execute
+ * @oparg:	argument of the operation
+ * @oval:	previous value at @uaddr on successful return
+ * @uaddr:	pointer to user space address
+ *
+ * Return:
+ * 0 - On success
+ * -EFAULT - User access resulted in a page fault
+ * -EAGAIN - Atomic operation was unable to complete due to contention
+ * -ENOSYS - Operation not supported
+ */
+static inline int
+futex_atomic_op_inuser_local(int op, u32 oparg, int *oval, u32 __user *uaddr)
+{
+	int oldval, ret;
+	u32 tmp;
+
+	preempt_disable();
+
+	ret = -EFAULT;
+	if (unlikely(get_user(oldval, uaddr) != 0))
+		goto out_pagefault_enable;
+
+	ret = 0;
+	tmp = oldval;
+
+	switch (op) {
+	case FUTEX_OP_SET:
+		tmp = oparg;
+		break;
+	case FUTEX_OP_ADD:
+		tmp += oparg;
+		break;
+	case FUTEX_OP_OR:
+		tmp |= oparg;
+		break;
+	case FUTEX_OP_ANDN:
+		tmp &= ~oparg;
+		break;
+	case FUTEX_OP_XOR:
+		tmp ^= oparg;
+		break;
+	default:
+		ret = -ENOSYS;
+	}
+
+	if (ret == 0 && unlikely(put_user(tmp, uaddr) != 0))
+		ret = -EFAULT;
+
+out_pagefault_enable:
+	preempt_enable();
+
+	if (ret == 0)
+		*oval = oldval;
+
+	return ret;
+}
+
+/**
+ * futex_atomic_cmpxchg_inatomic_local() - Compare and exchange the content of the
+ *				uaddr with newval if the current value is
+ *				oldval.
+ * @uval:	pointer to store content of @uaddr
+ * @uaddr:	pointer to user space address
+ * @oldval:	old value
+ * @newval:	new value to store to @uaddr
+ *
+ * Return:
+ * 0 - On success
+ * -EFAULT - User access resulted in a page fault
+ * -EAGAIN - Atomic operation was unable to complete due to contention
+ */
+static inline int
+futex_atomic_cmpxchg_inatomic_local(u32 *uval, u32 __user *uaddr,
+			      u32 oldval, u32 newval)
+{
+	u32 val;
+
+	preempt_disable();
+	if (unlikely(get_user(val, uaddr) != 0)) {
+		preempt_enable();
+		return -EFAULT;
+	}
+
+	if (val == oldval && unlikely(put_user(newval, uaddr) != 0)) {
+		preempt_enable();
+		return -EFAULT;
+	}
+
+	*uval = val;
+	preempt_enable();
+
+	return 0;
+}
+
+#endif