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, 130 insertions, 0 deletions

diff --git a/scripts/update-intel-ucode-defs.py b/scripts/update-intel-ucode-defs.py
new file mode 100755
index 000000000..9d6cc2c60
--- /dev/null
+++ b/scripts/update-intel-ucode-defs.py
@@ -0,0 +1,130 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: GPL-2.0
+import argparse
+import re
+import shutil
+import subprocess
+import sys
+import os
+
+script = os.path.relpath(__file__)
+
+DESCRIPTION = f"""
+For Intel CPUs, update the microcode revisions that determine
+X86_BUG_OLD_MICROCODE.
+
+This script is intended to be run in response to releases of the
+official Intel microcode GitHub repository:
+https://github.com/intel/Intel-Linux-Processor-Microcode-Data-Files.git
+
+It takes the Intel microcode files as input and uses iucode-tool to
+extract the revision information. It prints the output in the format
+expected by intel-ucode-defs.h.
+
+Usage:
+    ./{script} /path/to/microcode/files > /path/to/intel-ucode-defs.h
+
+Typically, someone at Intel would see a new public release, wait for at
+least three months to ensure the update is stable, run this script to
+refresh the intel-ucode-defs.h file, and send a patch upstream to update
+the mainline and stable versions.
+
+Any exception to this process should be supported with an appropriate
+justification.
+"""
+
+SIG_RE = re.compile(r'sig (0x[0-9a-fA-F]+)')
+PFM_RE = re.compile(r'pf_mask (0x[0-9a-fA-F]+)')
+REV_RE = re.compile(r'rev (0x[0-9a-fA-F]+)')
+
+# Functions to extract family, model, and stepping
+def bits(val, bottom, top):
+    mask = (1 << (top + 1 - bottom)) - 1
+    return (val >> bottom) & mask
+
+def family(sig):
+    if bits(sig, 8, 11) == 0xf:
+        return bits(sig, 8, 11) + bits(sig, 20, 27)
+    return bits(sig, 8, 11)
+
+def model(sig):
+    return bits(sig, 4, 7) | (bits(sig, 16, 19) << 4)
+
+def step(sig):
+    return bits(sig, 0, 3)
+
+class Ucode:
+    def __init__(self, sig, pfm, rev):
+        self.family = family(sig)
+        self.model = model(sig)
+        self.steppings = 1 << step(sig)
+        self.platforms = pfm
+        self.rev = rev
+
+        self.key = (self.family, self.model, self.steppings, self.platforms)
+
+    def __eq__(self, other):
+        return self.key == other.key
+
+    def __hash__(self):
+        return hash(self.key)
+
+    def __str__(self):
+        return "{ .flags = X86_CPU_ID_FLAG_ENTRY_VALID, .vendor = X86_VENDOR_INTEL, .family = 0x%x, .model = 0x%02x, .steppings = 0x%04x, .platform_mask = 0x%02x, .driver_data = 0x%x }," % \
+                (self.family, self.model, self.steppings, self.platforms, self.rev)
+
+def main():
+    parser = argparse.ArgumentParser(description=DESCRIPTION,
+                                     formatter_class=argparse.RawDescriptionHelpFormatter)
+    parser.add_argument('ucode_files', nargs='+', help='Path(s) to the microcode files')
+
+    args = parser.parse_args()
+
+    # Process the microcode files using iucode-tool
+    iucode_tool = shutil.which("iucode-tool") or shutil.which("iucode_tool")
+    if iucode_tool is None:
+        print("Error: iucode-tool not found, please install it", file=sys.stderr)
+        sys.exit(1)
+
+    cmd = [iucode_tool, '--list-all'] + args.ucode_files
+
+    result = subprocess.run(cmd, capture_output=True, text=True)
+    if result.returncode != 0:
+        print("Error: iucode-tool ran into an error, exiting", file=sys.stderr)
+        if result.stderr:
+            print(result.stderr, file=sys.stderr, end='')
+        sys.exit(1)
+
+    ucodes = set()
+
+    # Parse the output of iucode-tool
+    for line in result.stdout.splitlines():
+        sig_match = SIG_RE.search(line)
+        pfm_match = PFM_RE.search(line)
+        rev_match = REV_RE.search(line)
+
+        if not (sig_match and pfm_match and rev_match):
+            continue
+
+        sig = int(sig_match.group(1), 16)
+        pfm = int(pfm_match.group(1), 16)
+        rev = int(rev_match.group(1), 16)
+        debug_rev = bits(rev, 31, 31)
+        if debug_rev != 0:
+            print("Error: Debug ucode file found, exiting", file=sys.stderr)
+            sys.exit(1)
+
+        ucodes.add(Ucode(sig, pfm, rev))
+
+    if not ucodes:
+        print("Error: No valid microcode files found, exiting", file=sys.stderr)
+        sys.exit(1)
+
+    # Sort and print the microcode entries
+    print("/* SPDX-License-Identifier: GPL-2.0 */")
+    print("/* Auto-generated by scripts/update-intel-ucode-defs.py */")
+    for u in sorted(ucodes, key=lambda x: x.key):
+        print(u)
+
+if __name__ == "__main__":
+    main()