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, 100 insertions, 0 deletions

diff --git a/drivers/usb/chipidea/udc.h b/drivers/usb/chipidea/udc.h
new file mode 100644
index 000000000..c8a47389a
--- /dev/null
+++ b/drivers/usb/chipidea/udc.h
@@ -0,0 +1,100 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * udc.h - ChipIdea UDC structures
+ *
+ * Copyright (C) 2008 Chipidea - MIPS Technologies, Inc. All rights reserved.
+ *
+ * Author: David Lopo
+ */
+
+#ifndef __DRIVERS_USB_CHIPIDEA_UDC_H
+#define __DRIVERS_USB_CHIPIDEA_UDC_H
+
+#include <linux/list.h>
+
+#define CTRL_PAYLOAD_MAX   64
+#define RX        0  /* similar to USB_DIR_OUT but can be used as an index */
+#define TX        1  /* similar to USB_DIR_IN  but can be used as an index */
+
+/* DMA layout of transfer descriptors */
+struct ci_hw_td {
+	/* 0 */
+	__le32 next;
+#define TD_TERMINATE          BIT(0)
+#define TD_ADDR_MASK          (0xFFFFFFEUL << 5)
+	/* 1 */
+	__le32 token;
+#define TD_STATUS             (0x00FFUL <<  0)
+#define TD_STATUS_TR_ERR      BIT(3)
+#define TD_STATUS_DT_ERR      BIT(5)
+#define TD_STATUS_HALTED      BIT(6)
+#define TD_STATUS_ACTIVE      BIT(7)
+#define TD_MULTO              (0x0003UL << 10)
+#define TD_IOC                BIT(15)
+#define TD_TOTAL_BYTES        (0x7FFFUL << 16)
+	/* 2 */
+	__le32 page[5];
+#define TD_CURR_OFFSET        (0x0FFFUL <<  0)
+#define TD_FRAME_NUM          (0x07FFUL <<  0)
+#define TD_RESERVED_MASK      (0x0FFFUL <<  0)
+} __attribute__ ((packed, aligned(4)));
+
+/* DMA layout of queue heads */
+struct ci_hw_qh {
+	/* 0 */
+	__le32 cap;
+#define QH_IOS                BIT(15)
+#define QH_MAX_PKT            (0x07FFUL << 16)
+#define QH_ZLT                BIT(29)
+#define QH_MULT               (0x0003UL << 30)
+#define QH_ISO_MULT(x)		((x >> 11) & 0x03)
+	/* 1 */
+	__le32 curr;
+	/* 2 - 8 */
+	struct ci_hw_td		td;
+	/* 9 */
+	__le32 RESERVED;
+	struct usb_ctrlrequest   setup;
+} __attribute__ ((packed, aligned(4)));
+
+struct td_node {
+	struct list_head	td;
+	dma_addr_t		dma;
+	struct ci_hw_td		*ptr;
+	int			td_remaining_size;
+};
+
+/**
+ * struct ci_hw_req - usb request representation
+ * @req: request structure for gadget drivers
+ * @queue: link to QH list
+ * @tds: link to TD list
+ * @sgt: hold original sglist when bounce sglist
+ */
+struct ci_hw_req {
+	struct usb_request	req;
+	struct list_head	queue;
+	struct list_head	tds;
+	struct sg_table		sgt;
+};
+
+#ifdef CONFIG_USB_CHIPIDEA_UDC
+
+int ci_hdrc_gadget_init(struct ci_hdrc *ci);
+void ci_hdrc_gadget_destroy(struct ci_hdrc *ci);
+
+#else
+
+static inline int ci_hdrc_gadget_init(struct ci_hdrc *ci)
+{
+	return -ENXIO;
+}
+
+static inline void ci_hdrc_gadget_destroy(struct ci_hdrc *ci)
+{
+
+}
+
+#endif
+
+#endif /* __DRIVERS_USB_CHIPIDEA_UDC_H */