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, 88 insertions, 0 deletions

diff --git a/Documentation/driver-api/hsi.rst b/Documentation/driver-api/hsi.rst
new file mode 100644
index 000000000..01b6bebfb
--- /dev/null
+++ b/Documentation/driver-api/hsi.rst
@@ -0,0 +1,88 @@
+High Speed Synchronous Serial Interface (HSI)
+=============================================
+
+Introduction
+---------------
+
+High Speed Synchronous Interface (HSI) is a full duplex, low latency protocol,
+that is optimized for die-level interconnect between an Application Processor
+and a Baseband chipset. It has been specified by the MIPI alliance in 2003 and
+implemented by multiple vendors since then.
+
+The HSI interface supports full duplex communication over multiple channels
+(typically 8) and is capable of reaching speeds up to 200 Mbit/s.
+
+The serial protocol uses two signals, DATA and FLAG as combined data and clock
+signals and an additional READY signal for flow control. An additional WAKE
+signal can be used to wakeup the chips from standby modes. The signals are
+commonly prefixed by AC for signals going from the application die to the
+cellular die and CA for signals going the other way around.
+
+::
+
+    +------------+                                 +---------------+
+    |  Cellular  |                                 |  Application  |
+    |    Die     |                                 |      Die      |
+    |            | - - - - - - CAWAKE - - - - - - >|               |
+    |           T|------------ CADATA ------------>|R              |
+    |           X|------------ CAFLAG ------------>|X              |
+    |            |<----------- ACREADY ------------|               |
+    |            |                                 |               |
+    |            |                                 |               |
+    |            |< - - - - -  ACWAKE - - - - - - -|               |
+    |           R|<----------- ACDATA -------------|T              |
+    |           X|<----------- ACFLAG -------------|X              |
+    |            |------------ CAREADY ----------->|               |
+    |            |                                 |               |
+    |            |                                 |               |
+    +------------+                                 +---------------+
+
+HSI Subsystem in Linux
+-------------------------
+
+In the Linux kernel the hsi subsystem is supposed to be used for HSI devices.
+The hsi subsystem contains drivers for hsi controllers including support for
+multi-port controllers and provides a generic API for using the HSI ports.
+
+It also contains HSI client drivers, which make use of the generic API to
+implement a protocol used on the HSI interface. These client drivers can
+use an arbitrary number of channels.
+
+hsi-char Device
+------------------
+
+Each port automatically registers a generic client driver called hsi_char,
+which provides a character device for userspace representing the HSI port.
+It can be used to communicate via HSI from userspace. Userspace may
+configure the hsi_char device using the following ioctl commands:
+
+HSC_RESET
+ flush the HSI port
+
+HSC_SET_PM
+ enable or disable the client.
+
+HSC_SEND_BREAK
+ send break
+
+HSC_SET_RX
+ set RX configuration
+
+HSC_GET_RX
+ get RX configuration
+
+HSC_SET_TX
+ set TX configuration
+
+HSC_GET_TX
+ get TX configuration
+
+The kernel HSI API
+------------------
+
+.. kernel-doc:: include/linux/hsi/hsi.h
+   :internal:
+
+.. kernel-doc:: drivers/hsi/hsi_core.c
+   :export:
+