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, 120 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-driver-ccp b/Documentation/ABI/testing/sysfs-driver-ccp
new file mode 100644
index 000000000..6ec74b9a2
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-driver-ccp
@@ -0,0 +1,120 @@
+What:		/sys/bus/pci/devices/<BDF>/fused_part
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/fused_part file reports
+		whether the CPU or APU has been fused to prevent tampering.
+		0: Not fused
+		1: Fused
+
+What:		/sys/bus/pci/devices/<BDF>/boot_integrity
+Date:		April 2026
+KernelVersion:	6.20
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/boot_integrity reports
+		whether the AMD CPU or APU is used for a hardware root of trust
+		during the boot process.
+		Possible values:
+		0: Not hardware root of trust.
+		1: Hardware root of trust
+
+		NOTE: Vendors may provide design specific alternative hardware
+		root of trust implementations.
+
+What:		/sys/bus/pci/devices/<BDF>/debug_lock_on
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/debug_lock_on reports
+		whether the AMD CPU or APU has been unlocked for debugging.
+		Possible values:
+		0: Not locked
+		1: Locked
+
+What:		/sys/bus/pci/devices/<BDF>/tsme_status
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/tsme_status file reports
+		the status of transparent secure memory encryption on AMD systems.
+		Possible values:
+		0: Not active
+		1: Active
+
+What:		/sys/bus/pci/devices/<BDF>/anti_rollback_status
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/anti_rollback_status file reports
+		whether the PSP is enforcing rollback protection.
+		Possible values:
+		0: Not enforcing
+		1: Enforcing
+
+What:		/sys/bus/pci/devices/<BDF>/rpmc_production_enabled
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/rpmc_production_enabled file reports
+		whether Replay Protected Monotonic Counter support has been enabled.
+		Possible values:
+		0: Not enabled
+		1: Enabled
+
+What:		/sys/bus/pci/devices/<BDF>/rpmc_spirom_available
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/rpmc_spirom_available file reports
+		whether an Replay Protected Monotonic Counter supported SPI is installed
+		on the system.
+		Possible values:
+		0: Not present
+		1: Present
+
+What:		/sys/bus/pci/devices/<BDF>/hsp_tpm_available
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/hsp_tpm_available file reports
+		whether the HSP TPM has been activated.
+		Possible values:
+		0: Not activated or present
+		1: Activated
+
+What:		/sys/bus/pci/devices/<BDF>/rom_armor_enforced
+Date:		June 2022
+KernelVersion:	5.19
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/rom_armor_enforced file reports
+		whether RomArmor SPI protection is enforced.
+		Possible values:
+		0: Not enforced
+		1: Enforced
+
+What:		/sys/bus/pci/devices/<BDF>/bootloader_version
+Date:		June 2023
+KernelVersion:	6.4
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/bootloader_version
+		file reports the firmware version of the AMD AGESA
+		bootloader.
+
+What:		/sys/bus/pci/devices/<BDF>/tee_version
+Date:		June 2023
+KernelVersion:	6.4
+Contact:	mario.limonciello@amd.com
+Description:
+		The /sys/bus/pci/devices/<BDF>/tee_version
+		file reports the firmware version of the AMD Trusted
+		Execution Environment (TEE).