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, 173 insertions, 0 deletions

diff --git a/Documentation/sphinx/kernel_abi.py b/Documentation/sphinx/kernel_abi.py
new file mode 100644
index 000000000..5667f207d
--- /dev/null
+++ b/Documentation/sphinx/kernel_abi.py
@@ -0,0 +1,173 @@
+# -*- coding: utf-8; mode: python -*-
+# coding=utf-8
+# SPDX-License-Identifier: GPL-2.0
+#
+"""
+    kernel-abi
+    ~~~~~~~~~~
+
+    Implementation of the ``kernel-abi`` reST-directive.
+
+    :copyright:  Copyright (C) 2016  Markus Heiser
+    :copyright:  Copyright (C) 2016-2020  Mauro Carvalho Chehab
+    :maintained-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
+    :license:    GPL Version 2, June 1991 see Linux/COPYING for details.
+
+    The ``kernel-abi`` (:py:class:`KernelCmd`) directive calls the
+    AbiParser class to parse the Kernel ABI files.
+
+    Overview of directive's argument and options.
+
+    .. code-block:: rst
+
+        .. kernel-abi:: <ABI directory location>
+            :debug:
+
+    The argument ``<ABI directory location>`` is required. It contains the
+    location of the ABI files to be parsed.
+
+    ``debug``
+      Inserts a code-block with the *raw* reST. Sometimes it is helpful to see
+      what reST is generated.
+
+"""
+
+import os
+import re
+import sys
+
+from docutils import nodes, statemachine
+from docutils.statemachine import ViewList
+from docutils.parsers.rst import directives, Directive
+from sphinx.util.docutils import switch_source_input
+from sphinx.util import logging
+
+srctree = os.path.abspath(os.environ["srctree"])
+sys.path.insert(0, os.path.join(srctree, "tools/lib/python"))
+
+from abi.abi_parser import AbiParser
+
+__version__ = "1.0"
+
+logger = logging.getLogger('kernel_abi')
+path = os.path.join(srctree, "Documentation/ABI")
+
+_kernel_abi = None
+
+def get_kernel_abi():
+    """
+    Initialize kernel_abi global var, if not initialized yet.
+
+    This is needed to avoid warnings during Sphinx module initialization.
+    """
+    global _kernel_abi
+
+    if not _kernel_abi:
+        # Parse ABI symbols only once
+        _kernel_abi = AbiParser(path, logger=logger)
+        _kernel_abi.parse_abi()
+        _kernel_abi.check_issues()
+
+    return _kernel_abi
+
+def setup(app):
+
+    app.add_directive("kernel-abi", KernelCmd)
+    return {
+        "version": __version__,
+        "parallel_read_safe": True,
+        "parallel_write_safe": True
+    }
+
+
+class KernelCmd(Directive):
+    """KernelABI (``kernel-abi``) directive"""
+
+    required_arguments = 1
+    optional_arguments = 3
+    has_content = False
+    final_argument_whitespace = True
+    parser = None
+
+    option_spec = {
+        "debug": directives.flag,
+        "no-symbols": directives.flag,
+        "no-files":  directives.flag,
+    }
+
+    def run(self):
+        kernel_abi = get_kernel_abi()
+
+        doc = self.state.document
+        if not doc.settings.file_insertion_enabled:
+            raise self.warning("docutils: file insertion disabled")
+
+        env = self.state.document.settings.env
+        content = ViewList()
+        node = nodes.section()
+
+        abi_type = self.arguments[0]
+
+        if "no-symbols" in self.options:
+            show_symbols = False
+        else:
+            show_symbols = True
+
+        if "no-files" in self.options:
+            show_file = False
+        else:
+            show_file = True
+
+        tab_width = self.options.get('tab-width',
+                                     self.state.document.settings.tab_width)
+
+        old_f = None
+        n = 0
+        n_sym = 0
+        for msg, f, ln in kernel_abi.doc(show_file=show_file,
+                                            show_symbols=show_symbols,
+                                            filter_path=abi_type):
+            n_sym += 1
+            msg_list = statemachine.string2lines(msg, tab_width,
+                                                 convert_whitespace=True)
+            if "debug" in self.options:
+                lines = [
+                    "", "",  ".. code-block:: rst",
+                    "    :linenos:", ""
+                ]
+                for m in msg_list:
+                    lines.append("    " + m)
+            else:
+                lines = msg_list
+
+            for line in lines:
+                # sphinx counts lines from 0
+                content.append(line, f, ln - 1)
+                n += 1
+
+            if f != old_f:
+                # Add the file to Sphinx build dependencies if the file exists
+                fname = os.path.join(srctree, f)
+                if os.path.isfile(fname):
+                    env.note_dependency(fname)
+
+                old_f = f
+
+            # Sphinx doesn't like to parse big messages. So, let's
+            # add content symbol by symbol
+            if content:
+                self.do_parse(content, node)
+                content = ViewList()
+
+        if show_symbols and not show_file:
+            logger.verbose("%s ABI: %i symbols (%i ReST lines)" % (abi_type, n_sym, n))
+        elif not show_symbols and show_file:
+            logger.verbose("%s ABI: %i files (%i ReST lines)" % (abi_type, n_sym, n))
+        else:
+            logger.verbose("%s ABI: %i data (%i ReST lines)" % (abi_type, n_sym, n))
+
+        return node.children
+
+    def do_parse(self, content, node):
+        with switch_source_input(self.state, content):
+            self.state.nested_parse(content, 0, node, match_titles=1)