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, 127 insertions, 0 deletions

diff --git a/sound/soc/sof/iomem-utils.c b/sound/soc/sof/iomem-utils.c
new file mode 100644
index 000000000..f6cb79082
--- /dev/null
+++ b/sound/soc/sof/iomem-utils.c
@@ -0,0 +1,127 @@
+// SPDX-License-Identifier: (GPL-2.0-only OR BSD-3-Clause)
+//
+// This file is provided under a dual BSD/GPLv2 license.  When using or
+// redistributing this file, you may do so under either license.
+//
+// Copyright(c) 2018-2022 Intel Corporation
+//
+// Author: Keyon Jie <yang.jie@linux.intel.com>
+//
+
+#include <linux/io-64-nonatomic-lo-hi.h>
+#include <linux/platform_device.h>
+#include <linux/unaligned.h>
+#include <sound/soc.h>
+#include <sound/sof.h>
+#include "sof-priv.h"
+#include "ops.h"
+
+/*
+ * Register IO
+ *
+ * The sof_io_xyz() wrappers are typically referenced in snd_sof_dsp_ops
+ * structures and cannot be inlined.
+ */
+
+void sof_io_write(struct snd_sof_dev *sdev, void __iomem *addr, u32 value)
+{
+	writel(value, addr);
+}
+EXPORT_SYMBOL(sof_io_write);
+
+u32 sof_io_read(struct snd_sof_dev *sdev, void __iomem *addr)
+{
+	return readl(addr);
+}
+EXPORT_SYMBOL(sof_io_read);
+
+void sof_io_write64(struct snd_sof_dev *sdev, void __iomem *addr, u64 value)
+{
+	writeq(value, addr);
+}
+EXPORT_SYMBOL(sof_io_write64);
+
+u64 sof_io_read64(struct snd_sof_dev *sdev, void __iomem *addr)
+{
+	return readq(addr);
+}
+EXPORT_SYMBOL(sof_io_read64);
+
+/*
+ * IPC Mailbox IO
+ */
+
+void sof_mailbox_write(struct snd_sof_dev *sdev, u32 offset,
+		       void *message, size_t bytes)
+{
+	void __iomem *dest = sdev->bar[sdev->mailbox_bar] + offset;
+
+	memcpy_toio(dest, message, bytes);
+}
+EXPORT_SYMBOL(sof_mailbox_write);
+
+void sof_mailbox_read(struct snd_sof_dev *sdev, u32 offset,
+		      void *message, size_t bytes)
+{
+	void __iomem *src = sdev->bar[sdev->mailbox_bar] + offset;
+
+	memcpy_fromio(message, src, bytes);
+}
+EXPORT_SYMBOL(sof_mailbox_read);
+
+/*
+ * Memory copy.
+ */
+
+int sof_block_write(struct snd_sof_dev *sdev, enum snd_sof_fw_blk_type blk_type,
+		    u32 offset, void *src, size_t size)
+{
+	int bar = snd_sof_dsp_get_bar_index(sdev, blk_type);
+	const u8 *src_byte = src;
+	void __iomem *dest;
+	u32 affected_mask;
+	u32 tmp;
+	int m, n;
+
+	if (bar < 0)
+		return bar;
+
+	dest = sdev->bar[bar] + offset;
+
+	m = size / 4;
+	n = size % 4;
+
+	/* __iowrite32_copy use 32bit size values so divide by 4 */
+	__iowrite32_copy(dest, src, m);
+
+	if (n) {
+		affected_mask = (1 << (8 * n)) - 1;
+
+		/* first read the 32bit data of dest, then change affected
+		 * bytes, and write back to dest. For unaffected bytes, it
+		 * should not be changed
+		 */
+		tmp = ioread32(dest + m * 4);
+		tmp &= ~affected_mask;
+
+		tmp |= *(u32 *)(src_byte + m * 4) & affected_mask;
+		iowrite32(tmp, dest + m * 4);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(sof_block_write);
+
+int sof_block_read(struct snd_sof_dev *sdev, enum snd_sof_fw_blk_type blk_type,
+		   u32 offset, void *dest, size_t size)
+{
+	int bar = snd_sof_dsp_get_bar_index(sdev, blk_type);
+
+	if (bar < 0)
+		return bar;
+
+	memcpy_fromio(dest, sdev->bar[bar] + offset, size);
+
+	return 0;
+}
+EXPORT_SYMBOL(sof_block_read);