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, 98 insertions, 0 deletions

diff --git a/Documentation/ABI/README b/Documentation/ABI/README
new file mode 100644
index 000000000..27c962e6a
--- /dev/null
+++ b/Documentation/ABI/README
@@ -0,0 +1,98 @@
+This part of the documentation inside Documentation/ABI directory
+attempts to document the ABI between the Linux kernel and
+userspace, and the relative stability of these interfaces.  Due to the
+everchanging nature of Linux, and the differing maturity levels, these
+interfaces should be used by userspace programs in different ways.
+
+We have four different levels of ABI stability, as shown by the four
+different subdirectories in this location.  Interfaces may change levels
+of stability according to the rules described below.
+
+The different levels of stability are:
+
+  stable/
+	This directory documents the interfaces that the developer has
+	defined to be stable.  Userspace programs are free to use these
+	interfaces with no restrictions, and backward compatibility for
+	them will be guaranteed for at least 2 years.  Most interfaces
+	(like syscalls) are expected to never change and always be
+	available.
+
+  testing/
+	This directory documents interfaces that are felt to be stable,
+	as the main development of this interface has been completed.
+	The interface can be changed to add new features, but the
+	current interface will not break by doing this, unless grave
+	errors or security problems are found in them.  Userspace
+	programs can start to rely on these interfaces, but they must be
+	aware of changes that can occur before these interfaces move to
+	be marked stable.  Programs that use these interfaces are
+	strongly encouraged to add their name to the description of
+	these interfaces, so that the kernel developers can easily
+	notify them if any changes occur (see the description of the
+	layout of the files below for details on how to do this.)
+
+  obsolete/
+	This directory documents interfaces that are still remaining in
+	the kernel, but are marked to be removed at some later point in
+	time.  The description of the interface will document the reason
+	why it is obsolete and when it can be expected to be removed.
+
+  removed/
+	This directory contains a list of the old interfaces that have
+	been removed from the kernel.
+
+Every file in these directories will contain the following information:
+
+What:		Short description of the interface
+Date:		Date created
+KernelVersion:	(Optional) Kernel version this feature first showed up in.
+		Note: git history often provides more accurate version
+		info, so this field may be omitted.
+Contact:	Primary contact for this interface (may be a mailing list)
+Description:	Long description of the interface and how to use it.
+Users:		All users of this interface who wish to be notified when
+		it changes.  This is very important for interfaces in
+		the "testing" stage, so that kernel developers can work
+		with userspace developers to ensure that things do not
+		break in ways that are unacceptable.  It is also
+		important to get feedback for these interfaces to make
+		sure they are working in a proper way and do not need to
+		be changed further.
+
+
+Note:
+   The fields should use a simple notation, compatible with ReST markup.
+   Also, the file **should not** have a top-level index, like::
+
+	===
+	foo
+	===
+
+How things move between levels:
+
+Interfaces in stable may move to obsolete, as long as the proper
+notification is given.
+
+Interfaces may be removed from obsolete and the kernel as long as the
+documented amount of time has gone by.
+
+Interfaces in the testing state can move to the stable state when the
+developers feel they are finished.  They cannot be removed from the
+kernel tree without going through the obsolete state first.
+
+It's up to the developer to place their interfaces in the category they
+wish for it to start out in.
+
+
+Notable bits of non-ABI, which should not under any circumstances be considered
+stable:
+
+- Kconfig.  Userspace should not rely on the presence or absence of any
+  particular Kconfig symbol, in /proc/config.gz, in the copy of .config
+  commonly installed to /boot, or in any invocation of the kernel build
+  process.
+
+- Kernel-internal symbols.  Do not rely on the presence, absence, location, or
+  type of any kernel symbol, either in System.map files or the kernel binary
+  itself.  See Documentation/process/stable-api-nonsense.rst.