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, 61 insertions, 0 deletions

diff --git a/Documentation/ABI/testing/sysfs-kernel-kexec-kdump b/Documentation/ABI/testing/sysfs-kernel-kexec-kdump
new file mode 100644
index 000000000..f59051b5d
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-kernel-kexec-kdump
@@ -0,0 +1,61 @@
+What:		/sys/kernel/kexec/*
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:
+		The /sys/kernel/kexec/* directory contains sysfs files
+		that provide information about the configuration status
+		of kexec and kdump.
+
+What:		/sys/kernel/kexec/loaded
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:	read only
+		Indicates whether a new kernel image has been loaded
+		into memory using the kexec system call. It shows 1 if
+		a kexec image is present and ready to boot, or 0 if none
+		is loaded.
+User:		kexec tools, kdump service
+
+What:		/sys/kernel/kexec/crash_loaded
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:	read only
+		Indicates whether a crash (kdump) kernel is currently
+		loaded into memory. It shows 1 if a crash kernel has been
+		successfully loaded for panic handling, or 0 if no crash
+		kernel is present.
+User:		Kexec tools, Kdump service
+
+What:		/sys/kernel/kexec/crash_size
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:	read/write
+		Shows the amount of memory reserved for loading the crash
+		(kdump) kernel. It reports the size, in bytes, of the
+		crash kernel area defined by the crashkernel= parameter.
+		This interface also allows reducing the crashkernel
+		reservation by writing a smaller value, and the reclaimed
+		space is added back to the system RAM.
+User:		Kdump service
+
+What:		/sys/kernel/kexec/crash_elfcorehdr_size
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:	read only
+		Indicates the preferred size of the memory buffer for the
+		ELF core header used by the crash (kdump) kernel. It defines
+		how much space is needed to hold metadata about the crashed
+		system, including CPU and memory information. This information
+		is used by the user space utility kexec to support updating the
+		in-kernel kdump image during hotplug operations.
+User:		Kexec tools
+
+What:		/sys/kernel/kexec/crash_cma_ranges
+Date:		Nov 2025
+Contact:	kexec@lists.infradead.org
+Description:	read only
+		Provides information about the memory ranges reserved from
+		the Contiguous Memory Allocator (CMA) area that are allocated
+		to the crash (kdump) kernel. It lists the start and end physical
+		addresses of CMA regions assigned for crashkernel use.
+User:		kdump service