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, 75 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-class-net-phydev b/Documentation/ABI/testing/sysfs-class-net-phydev
new file mode 100644
index 000000000..31615c59b
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-class-net-phydev
@@ -0,0 +1,75 @@
+What:		/sys/class/mdio_bus/<bus>/<device>/attached_dev
+Date:		May 2017
+KernelVersion:	4.13
+Contact:	netdev@vger.kernel.org
+Description:
+		Symbolic link to the network device this PHY device is
+		attached to.
+
+What:		/sys/class/mdio_bus/<bus>/<device>/phy_has_fixups
+Date:		February 2014
+KernelVersion:	3.15
+Contact:	netdev@vger.kernel.org
+Description:
+		This attribute contains the boolean value whether a given PHY
+		device has had any "fixup" workaround running on it, encoded as
+		a boolean. This information is provided to help troubleshooting
+		PHY configurations.
+
+What:		/sys/class/mdio_bus/<bus>/<device>/phy_id
+Date:		November 2012
+KernelVersion:	3.8
+Contact:	netdev@vger.kernel.org
+Description:
+		This attribute contains the 32-bit PHY Identifier as reported
+		by the device during bus enumeration, encoded in hexadecimal.
+		This ID is used to match the device with the appropriate
+		driver.
+
+What:		/sys/class/mdio_bus/<bus>/<device>/c45_phy_ids/mmd<n>_device_id
+Date:		June 2025
+KernelVersion:	6.17
+Contact:	netdev@vger.kernel.org
+Description:
+		This attribute contains the 32-bit PHY Identifier as reported
+		by the device during bus enumeration, encoded in hexadecimal.
+		These C45 IDs are used to match the device with the appropriate
+		driver. These files are invisible to the C22 device.
+
+What:		/sys/class/mdio_bus/<bus>/<device>/phy_interface
+Date:		February 2014
+KernelVersion:	3.15
+Contact:	netdev@vger.kernel.org
+Description:
+		This attribute contains the PHY interface as configured by the
+		Ethernet driver during bus enumeration, encoded in string.
+		This interface mode is used to configure the Ethernet MAC with the
+		appropriate mode for its data lines to the PHY hardware.
+
+		Possible values are:
+
+		<empty> (not available), mii, gmii, sgmii, tbi, rev-mii,
+		rmii, rgmii, rgmii-id, rgmii-rxid, rgmii-txid, rtbi, smii
+		xgmii, moca, qsgmii, trgmii, 1000base-x, 2500base-x, rxaui,
+		xaui, 10gbase-kr, unknown
+
+What:		/sys/class/mdio_bus/<bus>/<device>/phy_standalone
+Date:		May 2019
+KernelVersion:	5.3
+Contact:	netdev@vger.kernel.org
+Description:
+		Boolean value indicating whether the PHY device is used in
+		standalone mode, without a net_device associated, by PHYLINK.
+		Attribute created only when this is the case.
+
+What:		/sys/class/mdio_bus/<bus>/<device>/phy_dev_flags
+Date:		March 2021
+KernelVersion:	5.13
+Contact:	netdev@vger.kernel.org
+Description:
+		32-bit hexadecimal number representing a bit mask of the
+		configuration bits passed from the consumer of the PHY
+		(Ethernet MAC, switch, etc.) to the PHY driver. The flags are
+		only used internally by the kernel and their placement are
+		not meant to be stable across kernel versions. This is intended
+		for facilitating the debugging of PHY drivers.