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/drivers/net/fddi/Kconfig b/drivers/net/fddi/Kconfig
new file mode 100644
index 000000000..846bf41c2
--- /dev/null
+++ b/drivers/net/fddi/Kconfig
@@ -0,0 +1,73 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# FDDI network device configuration
+#
+
+config FDDI
+	tristate "FDDI driver support"
+	depends on PCI || EISA || TC
+	help
+	  Fiber Distributed Data Interface is a high speed local area network
+	  design; essentially a replacement for high speed Ethernet. FDDI can
+	  run over copper or fiber. If you are connected to such a network and
+	  want a driver for the FDDI card in your computer, say Y here (and
+	  then also Y to the driver for your FDDI card, below). Most people
+	  will say N.
+
+if FDDI
+
+config DEFZA
+	tristate "DEC FDDIcontroller 700/700-C (DEFZA-xx) support"
+	depends on FDDI && TC
+	help
+	  This is support for the DEC FDDIcontroller 700 (DEFZA-AA, fiber)
+	  and 700-C (DEFZA-CA, copper) TURBOchannel network cards which
+	  can connect you to a local FDDI network.
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called defza.  If unsure, say N.
+
+config DEFXX
+	tristate "Digital DEFTA/DEFEA/DEFPA adapter support"
+	depends on FDDI && (PCI || EISA || TC)
+	help
+	  This is support for the DIGITAL series of TURBOchannel (DEFTA),
+	  EISA (DEFEA) and PCI (DEFPA) controllers which can connect you
+	  to a local FDDI network.
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called defxx.  If unsure, say N.
+
+config SKFP
+	tristate "SysKonnect FDDI PCI support"
+	depends on FDDI && PCI
+	select BITREVERSE
+	help
+	  Say Y here if you have a SysKonnect FDDI PCI adapter.
+	  The following adapters are supported by this driver:
+	  - SK-5521 (SK-NET FDDI-UP)
+	  - SK-5522 (SK-NET FDDI-UP DAS)
+	  - SK-5541 (SK-NET FDDI-FP)
+	  - SK-5543 (SK-NET FDDI-LP)
+	  - SK-5544 (SK-NET FDDI-LP DAS)
+	  - SK-5821 (SK-NET FDDI-UP64)
+	  - SK-5822 (SK-NET FDDI-UP64 DAS)
+	  - SK-5841 (SK-NET FDDI-FP64)
+	  - SK-5843 (SK-NET FDDI-LP64)
+	  - SK-5844 (SK-NET FDDI-LP64 DAS)
+	  - Netelligent 100 FDDI DAS Fibre SC
+	  - Netelligent 100 FDDI SAS Fibre SC
+	  - Netelligent 100 FDDI DAS UTP
+	  - Netelligent 100 FDDI SAS UTP
+	  - Netelligent 100 FDDI SAS Fibre MIC
+
+	  Read <file:Documentation/networking/device_drivers/fddi/skfp.rst>
+	  for information about the driver.
+
+	  Questions concerning this driver can be addressed to:
+	  <linux@syskonnect.de>
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called skfp.  This is recommended.
+
+endif # FDDI