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, 84 insertions, 0 deletions

diff --git a/arch/arm64/lib/delay.c b/arch/arm64/lib/delay.c
new file mode 100644
index 000000000..e278e060e
--- /dev/null
+++ b/arch/arm64/lib/delay.c
@@ -0,0 +1,84 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Delay loops based on the OpenRISC implementation.
+ *
+ * Copyright (C) 2012 ARM Limited
+ *
+ * Author: Will Deacon <will.deacon@arm.com>
+ */
+
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/timex.h>
+
+#include <clocksource/arm_arch_timer.h>
+
+#define USECS_TO_CYCLES(time_usecs)			\
+	xloops_to_cycles((time_usecs) * 0x10C7UL)
+
+static inline unsigned long xloops_to_cycles(unsigned long xloops)
+{
+	return (xloops * loops_per_jiffy * HZ) >> 32;
+}
+
+/*
+ * Force the use of CNTVCT_EL0 in order to have the same base as WFxT.
+ * This avoids some annoying issues when CNTVOFF_EL2 is not reset 0 on a
+ * KVM host running at EL1 until we do a vcpu_put() on the vcpu. When
+ * running at EL2, the effective offset is always 0.
+ *
+ * Note that userspace cannot change the offset behind our back either,
+ * as the vcpu mutex is held as long as KVM_RUN is in progress.
+ */
+static cycles_t notrace __delay_cycles(void)
+{
+	guard(preempt_notrace)();
+	return __arch_counter_get_cntvct_stable();
+}
+
+void __delay(unsigned long cycles)
+{
+	cycles_t start = __delay_cycles();
+
+	if (alternative_has_cap_unlikely(ARM64_HAS_WFXT)) {
+		u64 end = start + cycles;
+
+		/*
+		 * Start with WFIT. If an interrupt makes us resume
+		 * early, use a WFET loop to complete the delay.
+		 */
+		wfit(end);
+		while ((__delay_cycles() - start) < cycles)
+			wfet(end);
+	} else 	if (arch_timer_evtstrm_available()) {
+		const cycles_t timer_evt_period =
+			USECS_TO_CYCLES(ARCH_TIMER_EVT_STREAM_PERIOD_US);
+
+		while ((__delay_cycles() - start + timer_evt_period) < cycles)
+			wfe();
+	}
+
+	while ((__delay_cycles() - start) < cycles)
+		cpu_relax();
+}
+EXPORT_SYMBOL(__delay);
+
+inline void __const_udelay(unsigned long xloops)
+{
+	__delay(xloops_to_cycles(xloops));
+}
+EXPORT_SYMBOL(__const_udelay);
+
+void __udelay(unsigned long usecs)
+{
+	__const_udelay(usecs * 0x10C7UL); /* 2**32 / 1000000 (rounded up) */
+}
+EXPORT_SYMBOL(__udelay);
+
+void __ndelay(unsigned long nsecs)
+{
+	__const_udelay(nsecs * 0x5UL); /* 2**32 / 1000000000 (rounded up) */
+}
+EXPORT_SYMBOL(__ndelay);