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, 117 insertions, 0 deletions

diff --git a/lib/kunit/user_alloc.c b/lib/kunit/user_alloc.c
new file mode 100644
index 000000000..b8cac765e
--- /dev/null
+++ b/lib/kunit/user_alloc.c
@@ -0,0 +1,117 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * KUnit userspace memory allocation resource management.
+ */
+#include <kunit/resource.h>
+#include <kunit/test.h>
+#include <linux/kthread.h>
+#include <linux/mm.h>
+
+struct kunit_vm_mmap_resource {
+	unsigned long addr;
+	size_t size;
+};
+
+/* vm_mmap() arguments */
+struct kunit_vm_mmap_params {
+	struct file *file;
+	unsigned long addr;
+	unsigned long len;
+	unsigned long prot;
+	unsigned long flag;
+	unsigned long offset;
+};
+
+int kunit_attach_mm(void)
+{
+	struct mm_struct *mm;
+
+	if (current->mm)
+		return 0;
+
+	/* arch_pick_mmap_layout() is only sane with MMU systems. */
+	if (!IS_ENABLED(CONFIG_MMU))
+		return -EINVAL;
+
+	mm = mm_alloc();
+	if (!mm)
+		return -ENOMEM;
+
+	/* Define the task size. */
+	mm->task_size = TASK_SIZE;
+
+	/* Make sure we can allocate new VMAs. */
+	arch_pick_mmap_layout(mm, &current->signal->rlim[RLIMIT_STACK]);
+
+	/* Attach the mm. It will be cleaned up when the process dies. */
+	kthread_use_mm(mm);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kunit_attach_mm);
+
+static int kunit_vm_mmap_init(struct kunit_resource *res, void *context)
+{
+	struct kunit_vm_mmap_params *p = context;
+	struct kunit_vm_mmap_resource vres;
+	int ret;
+
+	ret = kunit_attach_mm();
+	if (ret)
+		return ret;
+
+	vres.size = p->len;
+	vres.addr = vm_mmap(p->file, p->addr, p->len, p->prot, p->flag, p->offset);
+	if (!vres.addr)
+		return -ENOMEM;
+	res->data = kmemdup(&vres, sizeof(vres), GFP_KERNEL);
+	if (!res->data) {
+		vm_munmap(vres.addr, vres.size);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static void kunit_vm_mmap_free(struct kunit_resource *res)
+{
+	struct kunit_vm_mmap_resource *vres = res->data;
+
+	/*
+	 * Since this is executed from the test monitoring process,
+	 * the test's mm has already been torn down. We don't need
+	 * to run vm_munmap(vres->addr, vres->size), only clean up
+	 * the vres.
+	 */
+
+	kfree(vres);
+	res->data = NULL;
+}
+
+unsigned long kunit_vm_mmap(struct kunit *test, struct file *file,
+			    unsigned long addr, unsigned long len,
+			    unsigned long prot, unsigned long flag,
+			    unsigned long offset)
+{
+	struct kunit_vm_mmap_params params = {
+		.file = file,
+		.addr = addr,
+		.len = len,
+		.prot = prot,
+		.flag = flag,
+		.offset = offset,
+	};
+	struct kunit_vm_mmap_resource *vres;
+
+	vres = kunit_alloc_resource(test,
+				    kunit_vm_mmap_init,
+				    kunit_vm_mmap_free,
+				    GFP_KERNEL,
+				    &params);
+	if (vres)
+		return vres->addr;
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kunit_vm_mmap);
+
+MODULE_IMPORT_NS("EXPORTED_FOR_KUNIT_TESTING");