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, 121 insertions, 0 deletions

diff --git a/include/trace/events/sof.h b/include/trace/events/sof.h
new file mode 100644
index 000000000..3061423c0
--- /dev/null
+++ b/include/trace/events/sof.h
@@ -0,0 +1,121 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright(c) 2022 Intel Corporation
+ *
+ * Author: Noah Klayman <noah.klayman@intel.com>
+ */
+
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM sof
+
+#if !defined(_TRACE_SOF_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_SOF_H
+#include <linux/tracepoint.h>
+#include <linux/types.h>
+#include <sound/sof/stream.h>
+#include "../../../sound/soc/sof/sof-audio.h"
+
+DECLARE_EVENT_CLASS(sof_widget_template,
+	TP_PROTO(struct snd_sof_widget *swidget),
+	TP_ARGS(swidget),
+	TP_STRUCT__entry(
+		__string(name, swidget->widget->name)
+		__field(int, use_count)
+	),
+	TP_fast_assign(
+		__assign_str(name);
+		__entry->use_count = swidget->use_count;
+	),
+	TP_printk("name=%s use_count=%d", __get_str(name), __entry->use_count)
+);
+
+DEFINE_EVENT(sof_widget_template, sof_widget_setup,
+	TP_PROTO(struct snd_sof_widget *swidget),
+	TP_ARGS(swidget)
+);
+
+DEFINE_EVENT(sof_widget_template, sof_widget_free,
+	TP_PROTO(struct snd_sof_widget *swidget),
+	TP_ARGS(swidget)
+);
+
+TRACE_EVENT(sof_ipc3_period_elapsed_position,
+	TP_PROTO(struct snd_sof_dev *sdev, struct sof_ipc_stream_posn *posn),
+	TP_ARGS(sdev, posn),
+	TP_STRUCT__entry(
+		__string(device_name, dev_name(sdev->dev))
+		__field(u64, host_posn)
+		__field(u64, dai_posn)
+		__field(u64, wallclock)
+	),
+	TP_fast_assign(
+		__assign_str(device_name);
+		__entry->host_posn = posn->host_posn;
+		__entry->dai_posn = posn->dai_posn;
+		__entry->wallclock = posn->wallclock;
+	),
+	TP_printk("device_name=%s host_posn=%#llx dai_posn=%#llx wallclock=%#llx",
+		  __get_str(device_name), __entry->host_posn, __entry->dai_posn,
+		  __entry->wallclock)
+);
+
+TRACE_EVENT(sof_pcm_pointer_position,
+	TP_PROTO(struct snd_sof_dev *sdev,
+		struct snd_sof_pcm *spcm,
+		struct snd_pcm_substream *substream,
+		snd_pcm_uframes_t dma_posn,
+		snd_pcm_uframes_t dai_posn
+	),
+	TP_ARGS(sdev, spcm, substream, dma_posn, dai_posn),
+	TP_STRUCT__entry(
+		__string(device_name, dev_name(sdev->dev))
+		__field(u32, pcm_id)
+		__field(int, stream)
+		__field(unsigned long, dma_posn)
+		__field(unsigned long, dai_posn)
+	),
+	TP_fast_assign(
+		__assign_str(device_name);
+		__entry->pcm_id = le32_to_cpu(spcm->pcm.pcm_id);
+		__entry->stream = substream->stream;
+		__entry->dma_posn = dma_posn;
+		__entry->dai_posn = dai_posn;
+	),
+	TP_printk("device_name=%s pcm_id=%d stream=%d dma_posn=%lu dai_posn=%lu",
+		  __get_str(device_name), __entry->pcm_id, __entry->stream,
+		  __entry->dma_posn, __entry->dai_posn)
+);
+
+TRACE_EVENT(sof_stream_position_ipc_rx,
+	TP_PROTO(struct device *dev),
+	TP_ARGS(dev),
+	TP_STRUCT__entry(
+		__string(device_name, dev_name(dev))
+	),
+	TP_fast_assign(
+		__assign_str(device_name);
+	),
+	TP_printk("device_name=%s", __get_str(device_name))
+);
+
+TRACE_EVENT(sof_ipc4_fw_config,
+	TP_PROTO(struct snd_sof_dev *sdev, char *key, u32 value),
+	TP_ARGS(sdev, key, value),
+	TP_STRUCT__entry(
+		__string(device_name, dev_name(sdev->dev))
+		__string(key, key)
+		__field(u32, value)
+	),
+	TP_fast_assign(
+		__assign_str(device_name);
+		__assign_str(key);
+		__entry->value = value;
+	),
+	TP_printk("device_name=%s key=%s value=%d",
+		  __get_str(device_name), __get_str(key), __entry->value)
+);
+
+#endif /* _TRACE_SOF_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>