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, 123 insertions, 0 deletions

diff --git a/drivers/phy/ti/Kconfig b/drivers/phy/ti/Kconfig
new file mode 100644
index 000000000..b52c7eabe
--- /dev/null
+++ b/drivers/phy/ti/Kconfig
@@ -0,0 +1,123 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Phy drivers for TI platforms
+#
+config OMAP_CONTROL_PHY
+	tristate "OMAP CONTROL PHY Driver"
+	depends on ARCH_OMAP2PLUS || COMPILE_TEST
+	help
+	  Enable this to add support for the PHY part present in the control
+	  module. This driver has API to power on the USB2 PHY and to write to
+	  the mailbox. The mailbox is present only in omap4 and the register to
+	  power on the USB2 PHY is present in OMAP4 and OMAP5. OMAP5 has an
+	  additional register to power on USB3 PHY/SATA PHY/PCIE PHY
+	  (PIPE3 PHY).
+
+config OMAP_USB2
+	tristate "OMAP USB2 PHY Driver"
+	depends on ARCH_OMAP2PLUS || ARCH_K3 || COMPILE_TEST
+	depends on USB_SUPPORT
+	select GENERIC_PHY
+	select USB_PHY
+	select OMAP_CONTROL_PHY if ARCH_OMAP2PLUS || COMPILE_TEST
+	help
+	  Enable this to support the transceiver that is part of SOC. This
+	  driver takes care of all the PHY functionality apart from comparator.
+	  The USB OTG controller communicates with the comparator using this
+	  driver.
+
+config PHY_AM654_SERDES
+	tristate "TI AM654 SERDES support"
+	depends on OF && (ARCH_K3 || COMPILE_TEST)
+	depends on COMMON_CLK
+	select GENERIC_PHY
+	select MULTIPLEXER
+	select REGMAP_MMIO
+	select MUX_MMIO
+	help
+	  This option enables support for TI AM654 SerDes PHY used for
+	  PCIe.
+
+config PHY_DA8XX_USB
+	tristate "TI DA8xx USB PHY Driver"
+	depends on ARCH_DAVINCI_DA8XX || COMPILE_TEST
+	select GENERIC_PHY
+	select MFD_SYSCON
+	help
+	  Enable this to support the USB PHY on DA8xx SoCs.
+
+	  This driver controls both the USB 1.1 PHY and the USB 2.0 PHY.
+
+config PHY_DM816X_USB
+	tristate "TI dm816x USB PHY driver"
+	depends on ARCH_OMAP2PLUS || COMPILE_TEST
+	depends on USB_SUPPORT
+	select GENERIC_PHY
+	select USB_PHY
+	help
+	  Enable this for dm816x USB to work.
+
+config PHY_J721E_WIZ
+	tristate "TI J721E WIZ (SERDES Wrapper) support"
+	depends on OF && (ARCH_K3 || COMPILE_TEST)
+	depends on HAS_IOMEM && OF_ADDRESS
+	depends on COMMON_CLK
+	select GENERIC_PHY
+	select MULTIPLEXER
+	select REGMAP_MMIO
+	select MUX_MMIO
+	help
+	  This option enables support for WIZ module present in TI's J721E
+	  SoC. WIZ is a serdes wrapper used to configure some of the input
+	  signals to the SERDES (Sierra/Torrent). This driver configures
+	  three clock selects (pll0, pll1, dig) and resets for each of the
+	  lanes.
+
+config PHY_TI_DS125DF111
+	tristate "TI DS125DF111 2-Channel Retimer Driver"
+	depends on OF && I2C
+	select GENERIC_PHY
+	help
+	  Enable this to add support for configuration and runtime management
+	  of the TI DS125DF111 Multi-Protocol 2-Channel Retimer.
+	  The retimer is modeled as a Generic PHY and supports both 10G and 1G
+	  link speeds.
+
+config PHY_TI_GMII_SEL
+	tristate
+	select GENERIC_PHY
+	select REGMAP
+	help
+	  This driver supports configuring of the TI CPSW Port mode depending on
+	  the Ethernet PHY connected to the CPSW Port.
+
+config PHY_TUSB1210
+	tristate "TI TUSB1210 ULPI PHY module"
+	depends on USB_ULPI_BUS
+	select GENERIC_PHY
+	help
+	  Support for TI TUSB1210 USB ULPI PHY.
+
+config TI_PIPE3
+	tristate "TI PIPE3 PHY Driver"
+	depends on ARCH_OMAP2PLUS || COMPILE_TEST
+	select GENERIC_PHY
+	select OMAP_CONTROL_PHY
+	help
+	  Enable this to support the PIPE3 PHY that is part of TI SOCs. This
+	  driver takes care of all the PHY functionality apart from comparator.
+	  This driver interacts with the "OMAP Control PHY Driver" to power
+	  on/off the PHY.
+
+config TWL4030_USB
+	tristate "TWL4030 USB Transceiver Driver"
+	depends on TWL4030_CORE && REGULATOR_TWL4030 && USB_MUSB_OMAP2PLUS
+	depends on USB_SUPPORT
+	depends on USB_GADGET || !USB_GADGET # if USB_GADGET=m, this can't 'y'
+	select GENERIC_PHY
+	select USB_PHY
+	help
+	  Enable this to support the USB OTG transceiver on TWL4030
+	  family chips (including the TWL5030 and TPS659x0 devices).
+	  This transceiver supports high and full speed devices plus,
+	  in host mode, low speed.