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, 208 insertions, 0 deletions

diff --git a/tools/bootconfig/test-bootconfig.sh b/tools/bootconfig/test-bootconfig.sh
new file mode 100755
index 000000000..fc69f815c
--- /dev/null
+++ b/tools/bootconfig/test-bootconfig.sh
@@ -0,0 +1,208 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0-only
+
+echo "Boot config test script"
+
+if [ -d "$1" ]; then
+  TESTDIR=$1
+else
+  TESTDIR=.
+fi
+BOOTCONF=${TESTDIR}/bootconfig
+ALIGN=4
+
+INITRD=`mktemp ${TESTDIR}/initrd-XXXX`
+TEMPCONF=`mktemp ${TESTDIR}/temp-XXXX.bconf`
+OUTFILE=`mktemp ${TESTDIR}/tempout-XXXX`
+NG=0
+
+cleanup() {
+  rm -f $INITRD $TEMPCONF $OUTFILE
+  exit $NG
+}
+
+trap cleanup EXIT TERM
+
+NO=1
+
+xpass() { # pass test command
+  echo "test case $NO ($*)... "
+  if ! ($@ && printf "\t\t[OK]\n"); then
+     printf "\t\t[NG]\n"; NG=$((NG + 1))
+  fi
+  NO=$((NO + 1))
+}
+
+xfail() { # fail test command
+  echo "test case $NO ($*)... "
+  if ! (! $@ && printf "\t\t[OK]\n"); then
+     printf "\t\t[NG]\n"; NG=$((NG + 1))
+  fi
+  NO=$((NO + 1))
+}
+
+echo "Basic command test"
+xpass $BOOTCONF $INITRD
+
+echo "Delete command should success without bootconfig"
+xpass $BOOTCONF -d $INITRD
+
+dd if=/dev/zero of=$INITRD bs=4096 count=1
+printf "key = value;" > $TEMPCONF
+bconf_size=$(wc -c < $TEMPCONF)
+initrd_size=$(wc -c < $INITRD)
+
+echo "Apply command test"
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+new_size=$(wc -c < $INITRD)
+
+echo "Show command test"
+xpass $BOOTCONF $INITRD
+
+echo "File size check"
+total_size=$(expr $bconf_size + $initrd_size + 9 + 12 + $ALIGN - 1 )
+total_size=$(expr $total_size / $ALIGN)
+total_size=$(expr $total_size \* $ALIGN)
+xpass test $new_size -eq $total_size
+
+echo "Apply command repeat test"
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+
+echo "File size check"
+xpass test $new_size -eq $(wc -c < $INITRD)
+
+echo "Delete command check"
+xpass $BOOTCONF -d $INITRD
+
+echo "File size check"
+new_size=$(wc -c < $INITRD)
+xpass test $new_size -eq $initrd_size
+
+echo "No error messge while applying"
+dd if=/dev/zero of=$INITRD bs=4096 count=1
+printf " \0\0\0 \0\0\0" >> $INITRD
+$BOOTCONF -a $TEMPCONF $INITRD > $OUTFILE 2>&1
+xfail grep -i "failed" $OUTFILE
+xfail grep -i "error" $OUTFILE
+
+echo "Max node number check"
+
+awk '
+BEGIN {
+  for (i = 0; i < 26; i += 1)
+      printf("%c\n", 65 + i % 26)
+  for (i = 26; i < 8192; i += 1)
+      printf("%c%c%c\n", 65 + i % 26, 65 + (i / 26) % 26, 65 + (i / 26 / 26))
+}
+' > $TEMPCONF
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+
+printf "badnode\n" >> $TEMPCONF
+xfail $BOOTCONF -a $TEMPCONF $INITRD
+
+echo "Max filesize check"
+
+# Max size is 32767 (including terminal byte)
+printf "data = \"" > $TEMPCONF
+dd if=/dev/urandom bs=768 count=32 | base64 -w0 >> $TEMPCONF
+printf "\"\n" >> $TEMPCONF
+xfail $BOOTCONF -a $TEMPCONF $INITRD
+
+dd if=$TEMPCONF of=$OUTFILE bs=1 count=32764
+cp $OUTFILE $TEMPCONF
+printf "\"\n" >> $TEMPCONF	# add 2 bytes + terminal ('\"\n\0')
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+
+echo "Adding same-key values"
+cat > $TEMPCONF << EOF
+key = bar, baz
+key += qux
+EOF
+echo > $INITRD
+
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+$BOOTCONF $INITRD > $OUTFILE
+xpass grep -q "bar" $OUTFILE
+xpass grep -q "baz" $OUTFILE
+xpass grep -q "qux" $OUTFILE
+
+echo "Override same-key values"
+cat > $TEMPCONF << EOF
+key = bar, baz
+key := qux
+EOF
+echo > $INITRD
+
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+$BOOTCONF $INITRD > $OUTFILE
+xfail grep -q "bar" $OUTFILE
+xfail grep -q "baz" $OUTFILE
+xpass grep -q "qux" $OUTFILE
+
+echo "Double/single quotes test"
+printf "key = '\"string\"';" > $TEMPCONF
+$BOOTCONF -a $TEMPCONF $INITRD
+$BOOTCONF $INITRD > $TEMPCONF
+cat $TEMPCONF
+xpass grep \'\"string\"\' $TEMPCONF
+
+echo "Repeat same-key tree"
+cat > $TEMPCONF << EOF
+foo
+bar
+foo { buz }
+EOF
+echo > $INITRD
+
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+$BOOTCONF $INITRD > $OUTFILE
+xpass grep -q "bar" $OUTFILE
+
+
+echo "Remove/keep tailing spaces"
+cat > $TEMPCONF << EOF
+foo = val     # comment
+bar = "val2 " # comment
+EOF
+echo > $INITRD
+
+xpass $BOOTCONF -a $TEMPCONF $INITRD
+$BOOTCONF $INITRD > $OUTFILE
+xfail grep -q 'val[[:space:]]' $OUTFILE
+xpass grep -q 'val2[[:space:]]' $OUTFILE
+
+echo "Showing correct line:column of no closing brace"
+cat > $TEMPCONF << EOF
+foo {
+bar {
+}
+EOF
+$BOOTCONF -a $TEMPCONF $INITRD 2> $OUTFILE
+xpass grep -q "1:1" $OUTFILE
+
+echo "=== expected failure cases ==="
+for i in samples/bad-* ; do
+  xfail $BOOTCONF -a $i $INITRD
+done
+
+echo "=== expected success cases ==="
+for i in samples/good-* ; do
+  xpass $BOOTCONF -a $i $INITRD
+  x="samples/exp-"`basename $i`
+  $BOOTCONF $i > $TEMPCONF
+  xpass diff $x $TEMPCONF
+done
+
+
+echo
+echo "=== Summary ==="
+echo "# of Passed: $(expr $NO - $NG - 1)"
+echo "# of Failed: $NG"
+
+echo
+if [ $NG -eq 0 ]; then
+	echo "All tests passed"
+else
+	echo "$NG tests failed"
+	exit 1
+fi