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, 137 insertions, 0 deletions

diff --git a/drivers/firmware/imx/misc.c b/drivers/firmware/imx/misc.c
new file mode 100644
index 000000000..d073cb3ce
--- /dev/null
+++ b/drivers/firmware/imx/misc.c
@@ -0,0 +1,137 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (C) 2016 Freescale Semiconductor, Inc.
+ * Copyright 2017~2018 NXP
+ *  Author: Dong Aisheng <aisheng.dong@nxp.com>
+ *
+ * File containing client-side RPC functions for the MISC service. These
+ * function are ported to clients that communicate to the SC.
+ *
+ */
+
+#include <linux/firmware/imx/svc/misc.h>
+
+struct imx_sc_msg_req_misc_set_ctrl {
+	struct imx_sc_rpc_msg hdr;
+	u32 ctrl;
+	u32 val;
+	u16 resource;
+} __packed __aligned(4);
+
+struct imx_sc_msg_req_cpu_start {
+	struct imx_sc_rpc_msg hdr;
+	u32 address_hi;
+	u32 address_lo;
+	u16 resource;
+	u8 enable;
+} __packed __aligned(4);
+
+struct imx_sc_msg_req_misc_get_ctrl {
+	struct imx_sc_rpc_msg hdr;
+	u32 ctrl;
+	u16 resource;
+} __packed __aligned(4);
+
+struct imx_sc_msg_resp_misc_get_ctrl {
+	struct imx_sc_rpc_msg hdr;
+	u32 val;
+} __packed __aligned(4);
+
+/*
+ * This function sets a miscellaneous control value.
+ *
+ * @param[in]     ipc         IPC handle
+ * @param[in]     resource    resource the control is associated with
+ * @param[in]     ctrl        control to change
+ * @param[in]     val         value to apply to the control
+ *
+ * @return Returns 0 for success and < 0 for errors.
+ */
+
+int imx_sc_misc_set_control(struct imx_sc_ipc *ipc, u32 resource,
+			    u8 ctrl, u32 val)
+{
+	struct imx_sc_msg_req_misc_set_ctrl msg;
+	struct imx_sc_rpc_msg *hdr = &msg.hdr;
+
+	hdr->ver = IMX_SC_RPC_VERSION;
+	hdr->svc = (uint8_t)IMX_SC_RPC_SVC_MISC;
+	hdr->func = (uint8_t)IMX_SC_MISC_FUNC_SET_CONTROL;
+	hdr->size = 4;
+
+	msg.ctrl = ctrl;
+	msg.val = val;
+	msg.resource = resource;
+
+	return imx_scu_call_rpc(ipc, &msg, true);
+}
+EXPORT_SYMBOL(imx_sc_misc_set_control);
+
+/*
+ * This function gets a miscellaneous control value.
+ *
+ * @param[in]     ipc         IPC handle
+ * @param[in]     resource    resource the control is associated with
+ * @param[in]     ctrl        control to get
+ * @param[out]    val         pointer to return the control value
+ *
+ * @return Returns 0 for success and < 0 for errors.
+ */
+
+int imx_sc_misc_get_control(struct imx_sc_ipc *ipc, u32 resource,
+			    u8 ctrl, u32 *val)
+{
+	struct imx_sc_msg_req_misc_get_ctrl msg;
+	struct imx_sc_msg_resp_misc_get_ctrl *resp;
+	struct imx_sc_rpc_msg *hdr = &msg.hdr;
+	int ret;
+
+	hdr->ver = IMX_SC_RPC_VERSION;
+	hdr->svc = (uint8_t)IMX_SC_RPC_SVC_MISC;
+	hdr->func = (uint8_t)IMX_SC_MISC_FUNC_GET_CONTROL;
+	hdr->size = 3;
+
+	msg.ctrl = ctrl;
+	msg.resource = resource;
+
+	ret = imx_scu_call_rpc(ipc, &msg, true);
+	if (ret)
+		return ret;
+
+	resp = (struct imx_sc_msg_resp_misc_get_ctrl *)&msg;
+	if (val != NULL)
+		*val = resp->val;
+
+	return 0;
+}
+EXPORT_SYMBOL(imx_sc_misc_get_control);
+
+/*
+ * This function starts/stops a CPU identified by @resource
+ *
+ * @param[in]     ipc         IPC handle
+ * @param[in]     resource    resource the control is associated with
+ * @param[in]     enable      true for start, false for stop
+ * @param[in]     phys_addr   initial instruction address to be executed
+ *
+ * @return Returns 0 for success and < 0 for errors.
+ */
+int imx_sc_pm_cpu_start(struct imx_sc_ipc *ipc, u32 resource,
+			bool enable, u64 phys_addr)
+{
+	struct imx_sc_msg_req_cpu_start msg;
+	struct imx_sc_rpc_msg *hdr = &msg.hdr;
+
+	hdr->ver = IMX_SC_RPC_VERSION;
+	hdr->svc = IMX_SC_RPC_SVC_PM;
+	hdr->func = IMX_SC_PM_FUNC_CPU_START;
+	hdr->size = 4;
+
+	msg.address_hi = phys_addr >> 32;
+	msg.address_lo = phys_addr;
+	msg.resource = resource;
+	msg.enable = enable;
+
+	return imx_scu_call_rpc(ipc, &msg, true);
+}
+EXPORT_SYMBOL(imx_sc_pm_cpu_start);