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, 57 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/debugfs-vfio b/Documentation/ABI/testing/debugfs-vfio
new file mode 100644
index 000000000..ed2f29c3a
--- /dev/null
+++ b/Documentation/ABI/testing/debugfs-vfio
@@ -0,0 +1,57 @@
+What:		/sys/kernel/debug/vfio
+Date:		December 2023
+KernelVersion:	6.8
+Contact:	Longfang Liu <liulongfang@huawei.com>
+Description:	This debugfs file directory is used for debugging
+		of vfio devices, it's a common directory for all vfio devices.
+		Vfio core will create a device subdirectory under this
+		directory.
+
+What:		/sys/kernel/debug/vfio/<device>/migration
+Date:		December 2023
+KernelVersion:	6.8
+Contact:	Longfang Liu <liulongfang@huawei.com>
+Description:	This debugfs file directory is used for debugging
+		of vfio devices that support live migration.
+		The debugfs of each vfio device that supports live migration
+		could be created under this directory.
+
+What:		/sys/kernel/debug/vfio/<device>/migration/state
+Date:		December 2023
+KernelVersion:	6.8
+Contact:	Longfang Liu <liulongfang@huawei.com>
+Description:	Read the live migration status of the vfio device.
+		The contents of the state file reflects the migration state
+		relative to those defined in the vfio_device_mig_state enum
+
+What:		/sys/kernel/debug/vfio/<device>/migration/features
+Date:		Oct 2025
+KernelVersion:	6.18
+Contact:	Cédric Le Goater <clg@redhat.com>
+Description:	Read the migration features of the vfio device.
+
+What:		/sys/kernel/debug/vfio/<device>/pci
+Date:		June 2026
+KernelVersion:	7.2
+Contact:	Alex Williamson <alex.williamson@nvidia.com>
+Description:	This debugfs file directory is used for debugging
+		VFIO PCI devices.
+
+What:		/sys/kernel/debug/vfio/<device>/pci/nointxmask
+Date:		June 2026
+KernelVersion:	7.2
+Contact:	Alex Williamson <alex.williamson@nvidia.com>
+Description:	Read the nointxmask policy latched for this device.  This
+		policy governs whether the device may use PCI 2.3 style
+		INTx masking when supported, reporting a value of "N", or
+		requires APIC level INTx masking, reporting a value of "Y".
+
+What:		/sys/kernel/debug/vfio/<device>/pci/disable_idle_d3
+Date:		June 2026
+KernelVersion:	7.2
+Contact:	Alex Williamson <alex.williamson@nvidia.com>
+Description:	Read the disable_idle_d3 policy latched for this device.  This
+		policy governs whether the device PM runtime usage count is
+		kept elevated while the device is bound to the driver and
+		unused, reporting a value of "Y", or decremented to allow the
+		device to enter a low power state, reporting a value of "N".