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, 88 insertions, 0 deletions

diff --git a/include/drm/drm_ras.h b/include/drm/drm_ras.h
new file mode 100644
index 000000000..0beede3dd
--- /dev/null
+++ b/include/drm/drm_ras.h
@@ -0,0 +1,88 @@
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright © 2026 Intel Corporation
+ */
+
+#ifndef __DRM_RAS_H__
+#define __DRM_RAS_H__
+
+#include <linux/types.h>
+
+#include <uapi/drm/drm_ras.h>
+
+/**
+ * struct drm_ras_node - A DRM RAS Node
+ */
+struct drm_ras_node {
+	/** @id: Unique identifier for the node. Dynamically assigned. */
+	u32 id;
+	/**
+	 * @device_name: Human-readable name of the device. Given by the driver.
+	 */
+	const char *device_name;
+	/** @node_name: Human-readable name of the node. Given by the driver. */
+	const char *node_name;
+	/** @type: Type of the node (enum drm_ras_node_type). */
+	enum drm_ras_node_type type;
+
+	/* Error-Counter Related Callback and Variables */
+
+	/** @error_counter_range: Range of valid Error IDs for this node. */
+	struct {
+		/** @first: First valid Error ID. */
+		u32 first;
+		/** @last: Last valid Error ID. Mandatory entry. */
+		u32 last;
+	} error_counter_range;
+
+	/**
+	 * @query_error_counter:
+	 *
+	 * This callback is used by drm-ras to query a specific error counter.
+	 * Used for input check and to iterate all error counters in a node.
+	 *
+	 * Driver should expect query_error_counter() to be called with
+	 * error_id from `error_counter_range.first` to
+	 * `error_counter_range.last`.
+	 *
+	 * The @query_error_counter is a mandatory callback for
+	 * error_counter_node.
+	 *
+	 * Returns: 0 on success,
+	 *          -ENOENT when error_id is not supported as an indication that
+	 *                  drm_ras should silently skip this entry. Used for
+	 *                  supporting non-contiguous error ranges.
+	 *                  Driver is responsible for maintaining the list of
+	 *                  supported error IDs in the range of first to last.
+	 *          Other negative values on errors that should terminate the
+	 *          netlink query.
+	 */
+	int (*query_error_counter)(struct drm_ras_node *node, u32 error_id,
+				   const char **name, u32 *val);
+
+	/**
+	 * @clear_error_counter:
+	 *
+	 * This callback is used by drm_ras to clear a specific error counter.
+	 * Driver should implement this callback to support clearing error counters
+	 * of a node.
+	 *
+	 * Returns: 0 on success, negative error code on failure.
+	 */
+	int (*clear_error_counter)(struct drm_ras_node *node, u32 error_id);
+
+	/** @priv: Driver private data */
+	void *priv;
+};
+
+struct drm_device;
+
+#if IS_ENABLED(CONFIG_DRM_RAS)
+int drm_ras_node_register(struct drm_ras_node *node);
+void drm_ras_node_unregister(struct drm_ras_node *node);
+#else
+static inline int drm_ras_node_register(struct drm_ras_node *node) { return 0; }
+static inline void drm_ras_node_unregister(struct drm_ras_node *node) { }
+#endif
+
+#endif