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, 76 insertions, 0 deletions

diff --git a/sound/soc/sof/sof-utils.c b/sound/soc/sof/sof-utils.c
new file mode 100644
index 000000000..f70089317
--- /dev/null
+++ b/sound/soc/sof/sof-utils.c
@@ -0,0 +1,76 @@
+// 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/unaligned.h>
+#include <linux/io-64-nonatomic-lo-hi.h>
+#include <linux/device.h>
+#include <sound/memalloc.h>
+#include <linux/module.h>
+#include "sof-utils.h"
+
+/*
+ * Generic buffer page table creation.
+ * Take the each physical page address and drop the least significant unused
+ * bits from each (based on PAGE_SIZE). Then pack valid page address bits
+ * into compressed page table.
+ */
+
+int snd_sof_create_page_table(struct device *dev,
+			      struct snd_dma_buffer *dmab,
+			      unsigned char *page_table, size_t size)
+{
+	int i, pages;
+
+	pages = snd_sgbuf_aligned_pages(size);
+
+	dev_dbg(dev, "generating page table for %p size 0x%zx pages %d\n",
+		dmab->area, size, pages);
+
+	for (i = 0; i < pages; i++) {
+		/*
+		 * The number of valid address bits for each page is 20.
+		 * idx determines the byte position within page_table
+		 * where the current page's address is stored
+		 * in the compressed page_table.
+		 * This can be calculated by multiplying the page number by 2.5.
+		 */
+		u32 idx = (5 * i) >> 1;
+		u32 pfn = snd_sgbuf_get_addr(dmab, i * PAGE_SIZE) >> PAGE_SHIFT;
+		u8 *pg_table;
+
+		pg_table = (u8 *)(page_table + idx);
+
+		/*
+		 * pagetable compression:
+		 * byte 0     byte 1     byte 2     byte 3     byte 4     byte 5
+		 * ___________pfn 0__________ __________pfn 1___________  _pfn 2...
+		 * .... ....  .... ....  .... ....  .... ....  .... ....  ....
+		 * It is created by:
+		 * 1. set current location to 0, PFN index i to 0
+		 * 2. put pfn[i] at current location in Little Endian byte order
+		 * 3. calculate an intermediate value as
+		 *    x = (pfn[i+1] << 4) | (pfn[i] & 0xf)
+		 * 4. put x at offset (current location + 2) in LE byte order
+		 * 5. increment current location by 5 bytes, increment i by 2
+		 * 6. continue to (2)
+		 */
+		if (i & 1)
+			put_unaligned_le32((pg_table[0] & 0xf) | pfn << 4,
+					   pg_table);
+		else
+			put_unaligned_le32(pfn, pg_table);
+	}
+
+	return pages;
+}
+EXPORT_SYMBOL(snd_sof_create_page_table);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("SOF utils");