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, 230 insertions, 0 deletions

diff --git a/include/drm/drm_of.h b/include/drm/drm_of.h
new file mode 100644
index 000000000..ebebed14c
--- /dev/null
+++ b/include/drm/drm_of.h
@@ -0,0 +1,230 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __DRM_OF_H__
+#define __DRM_OF_H__
+
+#include <linux/err.h>
+#include <linux/of_graph.h>
+#if IS_ENABLED(CONFIG_OF) && IS_ENABLED(CONFIG_DRM_PANEL_BRIDGE)
+#include <linux/of.h>
+#include <drm/drm_bridge.h>
+#endif
+
+struct component_master_ops;
+struct component_match;
+struct device;
+struct drm_device;
+struct drm_encoder;
+struct drm_panel;
+struct drm_bridge;
+struct device_node;
+struct mipi_dsi_device_info;
+struct mipi_dsi_host;
+
+enum drm_panel_orientation;
+
+/**
+ * enum drm_lvds_dual_link_pixels - Pixel order of an LVDS dual-link connection
+ * @DRM_LVDS_DUAL_LINK_EVEN_ODD_PIXELS: Even pixels are expected to be generated
+ *    from the first port, odd pixels from the second port
+ * @DRM_LVDS_DUAL_LINK_ODD_EVEN_PIXELS: Odd pixels are expected to be generated
+ *    from the first port, even pixels from the second port
+ */
+enum drm_lvds_dual_link_pixels {
+	DRM_LVDS_DUAL_LINK_EVEN_ODD_PIXELS = 0,
+	DRM_LVDS_DUAL_LINK_ODD_EVEN_PIXELS = 1,
+};
+
+#ifdef CONFIG_OF
+uint32_t drm_of_crtc_port_mask(struct drm_device *dev,
+			    struct device_node *port);
+uint32_t drm_of_find_possible_crtcs(struct drm_device *dev,
+				    struct device_node *port);
+void drm_of_component_match_add(struct device *master,
+				struct component_match **matchptr,
+				int (*compare)(struct device *, void *),
+				struct device_node *node);
+int drm_of_component_probe(struct device *dev,
+			   int (*compare_of)(struct device *, void *),
+			   const struct component_master_ops *m_ops);
+int drm_of_encoder_active_endpoint(struct device_node *node,
+				   struct drm_encoder *encoder,
+				   struct of_endpoint *endpoint);
+int drm_of_get_panel_orientation(const struct device_node *np,
+				 enum drm_panel_orientation *orientation);
+int drm_of_find_panel_or_bridge(const struct device_node *np,
+				int port, int endpoint,
+				struct drm_panel **panel,
+				struct drm_bridge **bridge);
+int drm_of_lvds_get_dual_link_pixel_order(const struct device_node *port1,
+					  const struct device_node *port2);
+int drm_of_lvds_get_dual_link_pixel_order_sink(struct device_node *port1,
+					       struct device_node *port2);
+int drm_of_lvds_get_data_mapping(const struct device_node *port);
+int drm_of_get_data_lanes_count(const struct device_node *endpoint,
+				const unsigned int min, const unsigned int max);
+int drm_of_get_data_lanes_count_ep(const struct device_node *port,
+				   int port_reg, int reg,
+				   const unsigned int min,
+				   const unsigned int max);
+int drm_of_get_data_lanes_count_remote(const struct device_node *port,
+				       int port_reg, int reg,
+				       const unsigned int min,
+				       const unsigned int max);
+#else
+static inline uint32_t drm_of_crtc_port_mask(struct drm_device *dev,
+					  struct device_node *port)
+{
+	return 0;
+}
+
+static inline uint32_t drm_of_find_possible_crtcs(struct drm_device *dev,
+						  struct device_node *port)
+{
+	return 0;
+}
+
+static inline void
+drm_of_component_match_add(struct device *master,
+			   struct component_match **matchptr,
+			   int (*compare)(struct device *, void *),
+			   struct device_node *node)
+{
+}
+
+static inline int
+drm_of_component_probe(struct device *dev,
+		       int (*compare_of)(struct device *, void *),
+		       const struct component_master_ops *m_ops)
+{
+	return -EINVAL;
+}
+
+static inline int drm_of_encoder_active_endpoint(struct device_node *node,
+						 struct drm_encoder *encoder,
+						 struct of_endpoint *endpoint)
+{
+	return -EINVAL;
+}
+
+static inline int drm_of_get_panel_orientation(const struct device_node *np,
+					       enum drm_panel_orientation *orientation)
+{
+	return -EINVAL;
+}
+
+static inline int drm_of_find_panel_or_bridge(const struct device_node *np,
+					      int port, int endpoint,
+					      struct drm_panel **panel,
+					      struct drm_bridge **bridge)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_lvds_get_dual_link_pixel_order(const struct device_node *port1,
+				      const struct device_node *port2)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_lvds_get_dual_link_pixel_order_sink(struct device_node *port1,
+					   struct device_node *port2)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_lvds_get_data_mapping(const struct device_node *port)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_get_data_lanes_count(const struct device_node *endpoint,
+			    const unsigned int min, const unsigned int max)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_get_data_lanes_count_ep(const struct device_node *port,
+			       int port_reg, int reg,
+			       const unsigned int min,
+			       const unsigned int max)
+{
+	return -EINVAL;
+}
+
+static inline int
+drm_of_get_data_lanes_count_remote(const struct device_node *port,
+				   int port_reg, int reg,
+				   const unsigned int min,
+				   const unsigned int max)
+{
+	return -EINVAL;
+}
+#endif
+
+#if IS_ENABLED(CONFIG_OF) && IS_ENABLED(CONFIG_DRM_MIPI_DSI)
+struct mipi_dsi_host *drm_of_get_dsi_bus(struct device *dev);
+#else
+static inline struct
+mipi_dsi_host *drm_of_get_dsi_bus(struct device *dev)
+{
+	return ERR_PTR(-EINVAL);
+}
+#endif /* CONFIG_OF && CONFIG_DRM_MIPI_DSI */
+
+/*
+ * drm_of_panel_bridge_remove - remove panel bridge
+ * @np: device tree node containing panel bridge output ports
+ *
+ * Remove the panel bridge of a given DT node's port and endpoint number
+ *
+ * Returns zero if successful, or one of the standard error codes if it fails.
+ */
+static inline int drm_of_panel_bridge_remove(const struct device_node *np,
+					     int port, int endpoint)
+{
+#if IS_ENABLED(CONFIG_OF) && IS_ENABLED(CONFIG_DRM_PANEL_BRIDGE)
+	struct drm_bridge *bridge;
+	struct device_node *remote;
+
+	remote = of_graph_get_remote_node(np, port, endpoint);
+	if (!remote)
+		return -ENODEV;
+
+	bridge = of_drm_find_and_get_bridge(remote);
+	drm_panel_bridge_remove(bridge);
+
+	drm_bridge_put(bridge);
+	of_node_put(remote);
+
+	return 0;
+#else
+	return -EINVAL;
+#endif
+}
+
+static inline int drm_of_encoder_active_endpoint_id(struct device_node *node,
+						    struct drm_encoder *encoder)
+{
+	struct of_endpoint endpoint;
+	int ret = drm_of_encoder_active_endpoint(node, encoder,
+						 &endpoint);
+
+	return ret ?: endpoint.id;
+}
+
+static inline int drm_of_encoder_active_port_id(struct device_node *node,
+						struct drm_encoder *encoder)
+{
+	struct of_endpoint endpoint;
+	int ret = drm_of_encoder_active_endpoint(node, encoder,
+						 &endpoint);
+
+	return ret ?: endpoint.port;
+}
+
+#endif /* __DRM_OF_H__ */