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, 75 insertions, 0 deletions

diff --git a/Documentation/driver-api/tty/index.rst b/Documentation/driver-api/tty/index.rst
new file mode 100644
index 000000000..c1ffe3d1e
--- /dev/null
+++ b/Documentation/driver-api/tty/index.rst
@@ -0,0 +1,75 @@
+.. SPDX-License-Identifier: GPL-2.0
+
+===
+TTY
+===
+
+Teletypewriter (TTY) layer takes care of all those serial devices. Including
+the virtual ones like pseudoterminal (PTY).
+
+TTY structures
+==============
+
+There are several major TTY structures. Every TTY device in a system has a
+corresponding struct tty_port. These devices are maintained by a TTY driver
+which is struct tty_driver. This structure describes the driver but also
+contains a reference to operations which could be performed on the TTYs. It is
+struct tty_operations. Then, upon open, a struct tty_struct is allocated and
+lives until the final close. During this time, several callbacks from struct
+tty_operations are invoked by the TTY layer.
+
+Every character received by the kernel (both from devices and users) is passed
+through a preselected :doc:`tty_ldisc` (in
+short ldisc; in C, struct tty_ldisc_ops). Its task is to transform characters
+as defined by a particular ldisc or by user too. The default one is n_tty,
+implementing echoes, signal handling, jobs control, special characters
+processing, and more. The transformed characters are passed further to
+user/device, depending on the source.
+
+In-detail description of the named TTY structures is in separate documents:
+
+.. toctree::
+   :maxdepth: 2
+
+   tty_driver
+   tty_port
+   tty_struct
+   tty_ldisc
+   tty_buffer
+   tty_ioctl
+   tty_internals
+   console
+
+Writing TTY Driver
+==================
+
+Before one starts writing a TTY driver, they must consider
+:doc:`Serial <../serial/driver>` and :doc:`USB Serial <../../usb/usb-serial>`
+layers first. Drivers for serial devices can often use one of these specific
+layers to implement a serial driver. Only special devices should be handled
+directly by the TTY Layer. If you are about to write such a driver, read on.
+
+A *typical* sequence a TTY driver performs is as follows:
+
+#. Allocate and register a TTY driver (module init)
+#. Create and register TTY devices as they are probed (probe function)
+#. Handle TTY operations and events like interrupts (TTY core invokes the
+   former, the device the latter)
+#. Remove devices as they are going away (remove function)
+#. Unregister and free the TTY driver (module exit)
+
+Steps regarding driver, i.e. 1., 3., and 5. are described in detail in
+:doc:`tty_driver`. For the other two (devices handling), look into
+:doc:`tty_port`.
+
+Other Documentation
+===================
+
+Miscellaneous documentation can be further found in these documents:
+
+.. toctree::
+   :maxdepth: 2
+
+   moxa-smartio
+   n_gsm
+   n_tty