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, 206 insertions, 0 deletions

diff --git a/include/trace/events/rtc.h b/include/trace/events/rtc.h
new file mode 100644
index 000000000..621333f1c
--- /dev/null
+++ b/include/trace/events/rtc.h
@@ -0,0 +1,206 @@
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM rtc
+
+#if !defined(_TRACE_RTC_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_RTC_H
+
+#include <linux/rtc.h>
+#include <linux/tracepoint.h>
+
+DECLARE_EVENT_CLASS(rtc_time_alarm_class,
+
+	TP_PROTO(time64_t secs, int err),
+
+	TP_ARGS(secs, err),
+
+	TP_STRUCT__entry(
+		__field(time64_t, secs)
+		__field(int, err)
+	),
+
+	TP_fast_assign(
+		__entry->secs = secs;
+		__entry->err = err;
+	),
+
+	TP_printk("UTC (%lld) (%d)",
+		  __entry->secs, __entry->err
+	)
+);
+
+DEFINE_EVENT(rtc_time_alarm_class, rtc_set_time,
+
+	TP_PROTO(time64_t secs, int err),
+
+	TP_ARGS(secs, err)
+);
+
+DEFINE_EVENT(rtc_time_alarm_class, rtc_read_time,
+
+	TP_PROTO(time64_t secs, int err),
+
+	TP_ARGS(secs, err)
+);
+
+DEFINE_EVENT(rtc_time_alarm_class, rtc_set_alarm,
+
+	TP_PROTO(time64_t secs, int err),
+
+	TP_ARGS(secs, err)
+);
+
+DEFINE_EVENT(rtc_time_alarm_class, rtc_read_alarm,
+
+	TP_PROTO(time64_t secs, int err),
+
+	TP_ARGS(secs, err)
+);
+
+TRACE_EVENT(rtc_irq_set_freq,
+
+	TP_PROTO(int freq, int err),
+
+	TP_ARGS(freq, err),
+
+	TP_STRUCT__entry(
+		__field(int, freq)
+		__field(int, err)
+	),
+
+	TP_fast_assign(
+		__entry->freq = freq;
+		__entry->err = err;
+	),
+
+	TP_printk("set RTC periodic IRQ frequency:%u (%d)",
+		  __entry->freq, __entry->err
+	)
+);
+
+TRACE_EVENT(rtc_irq_set_state,
+
+	TP_PROTO(int enabled, int err),
+
+	TP_ARGS(enabled, err),
+
+	TP_STRUCT__entry(
+		__field(int, enabled)
+		__field(int, err)
+	),
+
+	TP_fast_assign(
+		__entry->enabled = enabled;
+		__entry->err = err;
+	),
+
+	TP_printk("%s RTC 2^N Hz periodic IRQs (%d)",
+		  __entry->enabled ? "enable" : "disable",
+		  __entry->err
+	)
+);
+
+TRACE_EVENT(rtc_alarm_irq_enable,
+
+	TP_PROTO(unsigned int enabled, int err),
+
+	TP_ARGS(enabled, err),
+
+	TP_STRUCT__entry(
+		__field(unsigned int, enabled)
+		__field(int, err)
+	),
+
+	TP_fast_assign(
+		__entry->enabled = enabled;
+		__entry->err = err;
+	),
+
+	TP_printk("%s RTC alarm IRQ (%d)",
+		  __entry->enabled ? "enable" : "disable",
+		  __entry->err
+	)
+);
+
+DECLARE_EVENT_CLASS(rtc_offset_class,
+
+	TP_PROTO(long offset, int err),
+
+	TP_ARGS(offset, err),
+
+	TP_STRUCT__entry(
+		__field(long, offset)
+		__field(int, err)
+	),
+
+	TP_fast_assign(
+		__entry->offset = offset;
+		__entry->err = err;
+	),
+
+	TP_printk("RTC offset: %ld (%d)",
+		  __entry->offset, __entry->err
+	)
+);
+
+DEFINE_EVENT(rtc_offset_class, rtc_set_offset,
+
+	TP_PROTO(long offset, int err),
+
+	TP_ARGS(offset, err)
+);
+
+DEFINE_EVENT(rtc_offset_class, rtc_read_offset,
+
+	TP_PROTO(long offset, int err),
+
+	TP_ARGS(offset, err)
+);
+
+DECLARE_EVENT_CLASS(rtc_timer_class,
+
+	TP_PROTO(struct rtc_timer *timer),
+
+	TP_ARGS(timer),
+
+	TP_STRUCT__entry(
+		__field(struct rtc_timer *, timer)
+		__field(ktime_t, expires)
+		__field(ktime_t, period)
+	),
+
+	TP_fast_assign(
+		__entry->timer = timer;
+		__entry->expires = timer->node.expires;
+		__entry->period = timer->period;
+	),
+
+	TP_printk("RTC timer:(%p) expires:%lld period:%lld",
+		  __entry->timer, __entry->expires, __entry->period
+	)
+);
+
+DEFINE_EVENT(rtc_timer_class, rtc_timer_enqueue,
+
+	TP_PROTO(struct rtc_timer *timer),
+
+	TP_ARGS(timer)
+);
+
+DEFINE_EVENT(rtc_timer_class, rtc_timer_dequeue,
+
+	TP_PROTO(struct rtc_timer *timer),
+
+	TP_ARGS(timer)
+);
+
+DEFINE_EVENT(rtc_timer_class, rtc_timer_fired,
+
+	TP_PROTO(struct rtc_timer *timer),
+
+	TP_ARGS(timer)
+);
+
+#endif /* _TRACE_RTC_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>