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, 184 insertions, 0 deletions

diff --git a/drivers/rtc/rtc-imx-sc.c b/drivers/rtc/rtc-imx-sc.c
new file mode 100644
index 000000000..814d51664
--- /dev/null
+++ b/drivers/rtc/rtc-imx-sc.c
@@ -0,0 +1,184 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2018 NXP.
+ */
+
+#include <dt-bindings/firmware/imx/rsrc.h>
+#include <linux/arm-smccc.h>
+#include <linux/firmware/imx/sci.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/rtc.h>
+
+#define IMX_SC_TIMER_FUNC_GET_RTC_SEC1970	9
+#define IMX_SC_TIMER_FUNC_SET_RTC_ALARM		8
+#define IMX_SC_TIMER_FUNC_SET_RTC_TIME		6
+
+#define IMX_SIP_SRTC			0xC2000002
+#define IMX_SIP_SRTC_SET_TIME		0x0
+
+#define SC_IRQ_GROUP_RTC    2
+#define SC_IRQ_RTC          1
+
+static struct imx_sc_ipc *rtc_ipc_handle;
+static struct rtc_device *imx_sc_rtc;
+
+struct imx_sc_msg_timer_get_rtc_time {
+	struct imx_sc_rpc_msg hdr;
+	u32 time;
+} __packed;
+
+struct imx_sc_msg_timer_rtc_set_alarm {
+	struct imx_sc_rpc_msg hdr;
+	u16 year;
+	u8 mon;
+	u8 day;
+	u8 hour;
+	u8 min;
+	u8 sec;
+} __packed __aligned(4);
+
+static int imx_sc_rtc_read_time(struct device *dev, struct rtc_time *tm)
+{
+	struct imx_sc_msg_timer_get_rtc_time msg;
+	struct imx_sc_rpc_msg *hdr = &msg.hdr;
+	int ret;
+
+	hdr->ver = IMX_SC_RPC_VERSION;
+	hdr->svc = IMX_SC_RPC_SVC_TIMER;
+	hdr->func = IMX_SC_TIMER_FUNC_GET_RTC_SEC1970;
+	hdr->size = 1;
+
+	ret = imx_scu_call_rpc(rtc_ipc_handle, &msg, true);
+	if (ret) {
+		dev_err(dev, "read rtc time failed, ret %d\n", ret);
+		return ret;
+	}
+
+	rtc_time64_to_tm(msg.time, tm);
+
+	return 0;
+}
+
+static int imx_sc_rtc_set_time(struct device *dev, struct rtc_time *tm)
+{
+	struct arm_smccc_res res;
+
+	/* pack 2 time parameters into 1 register, 16 bits for each */
+	arm_smccc_smc(IMX_SIP_SRTC, IMX_SIP_SRTC_SET_TIME,
+		      ((tm->tm_year + 1900) << 16) | (tm->tm_mon + 1),
+		      (tm->tm_mday << 16) | tm->tm_hour,
+		      (tm->tm_min << 16) | tm->tm_sec,
+		      0, 0, 0, &res);
+
+	return res.a0;
+}
+
+static int imx_sc_rtc_alarm_irq_enable(struct device *dev, unsigned int enable)
+{
+	return imx_scu_irq_group_enable(SC_IRQ_GROUP_RTC, SC_IRQ_RTC, enable);
+}
+
+static int imx_sc_rtc_set_alarm(struct device *dev, struct rtc_wkalrm *alrm)
+{
+	struct imx_sc_msg_timer_rtc_set_alarm msg;
+	struct imx_sc_rpc_msg *hdr = &msg.hdr;
+	int ret;
+	struct rtc_time *alrm_tm = &alrm->time;
+
+	hdr->ver = IMX_SC_RPC_VERSION;
+	hdr->svc = IMX_SC_RPC_SVC_TIMER;
+	hdr->func = IMX_SC_TIMER_FUNC_SET_RTC_ALARM;
+	hdr->size = 3;
+
+	msg.year = alrm_tm->tm_year + 1900;
+	msg.mon = alrm_tm->tm_mon + 1;
+	msg.day = alrm_tm->tm_mday;
+	msg.hour = alrm_tm->tm_hour;
+	msg.min = alrm_tm->tm_min;
+	msg.sec = alrm_tm->tm_sec;
+
+	ret = imx_scu_call_rpc(rtc_ipc_handle, &msg, true);
+	if (ret) {
+		dev_err(dev, "set rtc alarm failed, ret %d\n", ret);
+		return ret;
+	}
+
+	ret = imx_sc_rtc_alarm_irq_enable(dev, alrm->enabled);
+	if (ret) {
+		dev_err(dev, "enable rtc alarm failed, ret %d\n", ret);
+		return ret;
+	}
+
+	return 0;
+}
+
+static const struct rtc_class_ops imx_sc_rtc_ops = {
+	.read_time = imx_sc_rtc_read_time,
+	.set_time = imx_sc_rtc_set_time,
+	.set_alarm = imx_sc_rtc_set_alarm,
+	.alarm_irq_enable = imx_sc_rtc_alarm_irq_enable,
+};
+
+static int imx_sc_rtc_alarm_notify(struct notifier_block *nb,
+					unsigned long event, void *group)
+{
+	/* ignore non-rtc irq */
+	if (!((event & SC_IRQ_RTC) && (*(u8 *)group == SC_IRQ_GROUP_RTC)))
+		return 0;
+
+	rtc_update_irq(imx_sc_rtc, 1, RTC_IRQF | RTC_AF);
+
+	return 0;
+}
+
+static struct notifier_block imx_sc_rtc_alarm_sc_notifier = {
+	.notifier_call = imx_sc_rtc_alarm_notify,
+};
+
+static int imx_sc_rtc_probe(struct platform_device *pdev)
+{
+	int ret;
+
+	ret = imx_scu_get_handle(&rtc_ipc_handle);
+	if (ret)
+		return ret;
+
+	device_init_wakeup(&pdev->dev, true);
+
+	imx_sc_rtc = devm_rtc_allocate_device(&pdev->dev);
+	if (IS_ERR(imx_sc_rtc))
+		return PTR_ERR(imx_sc_rtc);
+
+	imx_sc_rtc->ops = &imx_sc_rtc_ops;
+	imx_sc_rtc->range_min = 0;
+	imx_sc_rtc->range_max = U32_MAX;
+
+	ret = devm_rtc_register_device(imx_sc_rtc);
+	if (ret)
+		return ret;
+
+	imx_scu_irq_register_notifier(&imx_sc_rtc_alarm_sc_notifier);
+
+	return 0;
+}
+
+static const struct of_device_id imx_sc_dt_ids[] = {
+	{ .compatible = "fsl,imx8qxp-sc-rtc", },
+	{}
+};
+MODULE_DEVICE_TABLE(of, imx_sc_dt_ids);
+
+static struct platform_driver imx_sc_rtc_driver = {
+	.driver = {
+		.name	= "imx-sc-rtc",
+		.of_match_table = imx_sc_dt_ids,
+	},
+	.probe		= imx_sc_rtc_probe,
+};
+module_platform_driver(imx_sc_rtc_driver);
+
+MODULE_AUTHOR("Anson Huang <Anson.Huang@nxp.com>");
+MODULE_DESCRIPTION("NXP i.MX System Controller RTC Driver");
+MODULE_LICENSE("GPL");