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/phy/mediatek/Kconfig b/drivers/phy/mediatek/Kconfig
new file mode 100644
index 000000000..ba6461350
--- /dev/null
+++ b/drivers/phy/mediatek/Kconfig
@@ -0,0 +1,100 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Phy drivers for Mediatek devices
+#
+config PHY_MTK_PCIE
+	tristate "MediaTek PCIe-PHY Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Say 'Y' here to add support for MediaTek PCIe PHY driver.
+	  This driver create the basic PHY instance and provides initialize
+	  callback for PCIe GEN3 port, it supports software efuse
+	  initialization.
+
+config PHY_MTK_XFI_TPHY
+	tristate "MediaTek 10GE SerDes XFI T-PHY driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Say 'Y' here to add support for MediaTek XFI T-PHY driver.
+	  The driver provides access to the Ethernet SerDes T-PHY supporting
+	  1GE and 2.5GE modes via the LynxI PCS, and 5GE and 10GE modes
+	  via the USXGMII PCS found in MediaTek SoCs with 10G Ethernet.
+
+config PHY_MTK_TPHY
+	tristate "MediaTek T-PHY Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF && OF_ADDRESS
+	depends on HAS_IOMEM
+	select GENERIC_PHY
+	help
+	  Say 'Y' here to add support for MediaTek T-PHY driver,
+	  it supports multiple usb2.0, usb3.0 ports, PCIe and
+	  SATA, and meanwhile supports two version T-PHY which have
+	  different banks layout, the T-PHY with shared banks between
+	  multi-ports is first version, otherwise is second version,
+	  so you can easily distinguish them by banks layout.
+
+config PHY_MTK_UFS
+	tristate "MediaTek UFS M-PHY driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Support for UFS M-PHY on MediaTek chipsets.
+	  Enable this to provide vendor-specific probing,
+	  initialization, power on and power off flow of
+	  specified M-PHYs.
+
+config PHY_MTK_XSPHY
+	tristate "MediaTek XS-PHY Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF && OF_ADDRESS
+	depends on HAS_IOMEM
+	select GENERIC_PHY
+	help
+	  Enable this to support the SuperSpeedPlus XS-PHY transceiver for
+	  USB3.1 GEN2 controllers on MediaTek chips. The driver supports
+	  multiple USB2.0, USB3.1 GEN2 ports.
+
+config PHY_MTK_HDMI
+	tristate "MediaTek HDMI-PHY Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on COMMON_CLK
+	depends on OF
+	depends on REGULATOR
+	select GENERIC_PHY
+	help
+	  Support HDMI PHY for Mediatek SoCs.
+
+config PHY_MTK_MIPI_CSI_0_5
+	tristate "MediaTek MIPI CSI CD-PHY v0.5 Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Enable this to support the MIPI CSI CD-PHY receiver version 0.5.
+	  The driver supports multiple CSI cdphy ports simultaneously.
+
+	  To compile this driver as a module, choose M here: the
+	  module will be called phy-mtk-mipi-csi-0-5.
+
+config PHY_MTK_MIPI_DSI
+	tristate "MediaTek MIPI-DSI Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on COMMON_CLK
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Support MIPI DSI for Mediatek SoCs.
+
+config PHY_MTK_DP
+	tristate "MediaTek DP-PHY Driver"
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	depends on OF
+	select GENERIC_PHY
+	help
+	  Support DisplayPort PHY for MediaTek SoCs.