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, 72 insertions, 0 deletions

diff --git a/sound/firewire/lib.c b/sound/firewire/lib.c
new file mode 100644
index 000000000..654e1a605
--- /dev/null
+++ b/sound/firewire/lib.c
@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * miscellaneous helper functions
+ *
+ * Copyright (c) Clemens Ladisch <clemens@ladisch.de>
+ */
+
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/firewire.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include "lib.h"
+
+#define ERROR_RETRY_DELAY_MS	20
+
+/**
+ * snd_fw_transaction - send a request and wait for its completion
+ * @unit: the driver's unit on the target device
+ * @tcode: the transaction code
+ * @offset: the address in the target's address space
+ * @buffer: input/output data
+ * @length: length of @buffer
+ * @flags: use %FW_FIXED_GENERATION and add the generation value to attempt the
+ *         request only in that generation; use %FW_QUIET to suppress error
+ *         messages
+ *
+ * Submits an asynchronous request to the target device, and waits for the
+ * response.  The node ID and the current generation are derived from @unit.
+ * On a bus reset or an error, the transaction is retried a few times.
+ * Returns zero on success, or a negative error code.
+ */
+int snd_fw_transaction(struct fw_unit *unit, int tcode,
+		       u64 offset, void *buffer, size_t length,
+		       unsigned int flags)
+{
+	struct fw_device *device = fw_parent_device(unit);
+	int generation, rcode, tries = 0;
+
+	generation = flags & FW_GENERATION_MASK;
+	for (;;) {
+		if (!(flags & FW_FIXED_GENERATION)) {
+			generation = device->generation;
+			smp_rmb(); /* node_id vs. generation */
+		}
+		rcode = fw_run_transaction(device->card, tcode,
+					   device->node_id, generation,
+					   device->max_speed, offset,
+					   buffer, length);
+
+		if (rcode == RCODE_COMPLETE)
+			return 0;
+
+		if (rcode == RCODE_GENERATION && (flags & FW_FIXED_GENERATION))
+			return -EAGAIN;
+
+		if (rcode_is_permanent_error(rcode) || ++tries >= 3) {
+			if (!(flags & FW_QUIET))
+				dev_err(&unit->device,
+					"transaction failed: %s\n",
+					fw_rcode_string(rcode));
+			return -EIO;
+		}
+
+		msleep(ERROR_RETRY_DELAY_MS);
+	}
+}
+EXPORT_SYMBOL(snd_fw_transaction);
+
+MODULE_DESCRIPTION("FireWire audio helper functions");
+MODULE_AUTHOR("Clemens Ladisch <clemens@ladisch.de>");
+MODULE_LICENSE("GPL");