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, 73 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/amdpmf-interface b/Documentation/ABI/testing/amdpmf-interface
new file mode 100644
index 000000000..625f90a01
--- /dev/null
+++ b/Documentation/ABI/testing/amdpmf-interface
@@ -0,0 +1,73 @@
+What:		/dev/amdpmf_interface
+Date:		June 2026
+KernelVersion:	7.2
+Contact:	Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
+Description:
+		The AMD Platform Management Framework (PMF) util layer exposes a
+		minimal user-space interface via a character device for feature
+		discovery and metrics monitoring.
+
+		When CONFIG_AMD_PMF_UTIL_SUPPORT is enabled, the driver creates
+		a character device:
+
+		======================
+		/dev/amdpmf_interface
+		======================
+
+		The interface supports a single ioctl:
+
+		============================ =======================================
+		IOCTL			     Usage
+		IOCTL_AMD_PMF_POPULATE_DATA  User passes a struct amd_pmf_info with
+					     the size field set to sizeof(struct
+					     amd_pmf_info). The driver returns all
+					     available metrics and feature status
+					     in the structure.
+		============================ =======================================
+
+		struct amd_pmf_info layout:
+
+		======================= ========== ================================
+		Field			Type	   Description
+		size			__u64	   Structure size for versioning
+		features_supported	__u32	   Bitmask of supported features
+		platform_type		__u32	   Platform form factor orientation
+		power_source		__u32	   AC/DC power source
+		laptop_placement	__u32	   Device placement state
+		lid_state		__u32	   Lid open/closed status
+		user_presence		__u32	   User presence detection
+		slider_position		__u32	   Current power slider position
+		skin_temp		__s32	   Skin temperature (centidegrees)
+		gfx_busy		__u32	   Graphics workload percentage
+		ambient_light		__s32	   Ambient light sensor reading
+		avg_c0_residency	__u32	   Average C0 state residency
+		max_c0_residency	__u32	   Maximum C0 state residency
+		socket_power		__u32	   Socket power consumption
+		bios_input[10]		__u32	   Custom BIOS input parameters
+		bios_output[10]		__u32	   Custom BIOS output parameters
+		======================= ========== ================================
+
+		Feature Support Flags (features_supported bitmask):
+
+		=====================================    ====  =============================
+		Flag				         Bit   Description
+		AMD_PMF_FEAT_AUTO_MODE		         0     Auto Mode feature support
+		AMD_PMF_FEAT_STATIC_POWER_SLIDER         1     Static Power Slider support
+		AMD_PMF_FEAT_POLICY_BUILDER	         2     Policy Builder (Smart PC)
+		AMD_PMF_FEAT_DYNAMIC_POWER_SLIDER_AC     3     Dynamic slider on AC
+		AMD_PMF_FEAT_DYNAMIC_POWER_SLIDER_DC     4     Dynamic slider on DC
+		=====================================    ====  =============================
+
+		Return codes:
+
+		============= ============================================================
+		Return code   Description
+		0	      Success
+		EINVAL	      Invalid size or parameter
+		EFAULT	      copy_to_user/copy_from_user failures
+		ENODEV	      PMF device not available
+		ENOTTY	      Unknown ioctl command
+		============= ============================================================
+
+		User-space tools integrating with AMD PMF to discover capabilities and
+		monitor real-time metrics for thermal and power management validation.