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, 70 insertions, 0 deletions

diff --git a/Documentation/admin-guide/mono.rst b/Documentation/admin-guide/mono.rst
new file mode 100644
index 000000000..c6dab5680
--- /dev/null
+++ b/Documentation/admin-guide/mono.rst
@@ -0,0 +1,70 @@
+Mono(tm) Binary Kernel Support for Linux
+-----------------------------------------
+
+To configure Linux to automatically execute Mono-based .NET binaries
+(in the form of .exe files) without the need to use the mono CLR
+wrapper, you can use the BINFMT_MISC kernel support.
+
+This will allow you to execute Mono-based .NET binaries just like any
+other program after you have done the following:
+
+1) You MUST FIRST install the Mono CLR support, either by downloading
+   a binary package, a source tarball or by installing from Git. Binary
+   packages for several distributions can be found at:
+
+	https://www.mono-project.com/download/
+
+   Instructions for compiling Mono can be found at:
+
+	https://www.mono-project.com/docs/compiling-mono/linux/
+
+   Once the Mono CLR support has been installed, just check that
+   ``/usr/bin/mono`` (which could be located elsewhere, for example
+   ``/usr/local/bin/mono``) is working.
+
+2) You have to compile BINFMT_MISC either as a module or into
+   the kernel (``CONFIG_BINFMT_MISC``) and set it up properly.
+   If you choose to compile it as a module, you will have
+   to insert it manually with modprobe/insmod, as kmod
+   cannot be easily supported with binfmt_misc.
+   Read the file ``binfmt_misc.txt`` in this directory to know
+   more about the configuration process.
+
+3) Add the following entries to ``/etc/rc.local`` or similar script
+   to be run at system startup:
+
+   .. code-block:: sh
+
+    # Insert BINFMT_MISC module into the kernel
+    if [ ! -e /proc/sys/fs/binfmt_misc/register ]; then
+        /sbin/modprobe binfmt_misc
+	# Some distributions, like Fedora Core, perform
+	# the following command automatically when the
+	# binfmt_misc module is loaded into the kernel
+	# or during normal boot up (systemd-based systems).
+	# Thus, it is possible that the following line
+	# is not needed at all.
+	mount -t binfmt_misc none /proc/sys/fs/binfmt_misc
+    fi
+
+    # Register support for .NET CLR binaries
+    if [ -e /proc/sys/fs/binfmt_misc/register ]; then
+	# Replace /usr/bin/mono with the correct pathname to
+	# the Mono CLR runtime (usually /usr/local/bin/mono
+	# when compiling from sources or CVS).
+        echo ':CLR:M::MZ::/usr/bin/mono:' > /proc/sys/fs/binfmt_misc/register
+    else
+        echo "No binfmt_misc support"
+        exit 1
+    fi
+
+4) Check that ``.exe`` binaries can be ran without the need of a
+   wrapper script, simply by launching the ``.exe`` file directly
+   from a command prompt, for example::
+
+	/usr/bin/xsd.exe
+
+   .. note::
+
+      If this fails with a permission denied error, check
+      that the ``.exe`` file has execute permissions.