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, 225 insertions, 0 deletions

diff --git a/mm/page_idle.c b/mm/page_idle.c
new file mode 100644
index 000000000..9c67cbac2
--- /dev/null
+++ b/mm/page_idle.c
@@ -0,0 +1,225 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/init.h>
+#include <linux/memblock.h>
+#include <linux/fs.h>
+#include <linux/sysfs.h>
+#include <linux/kobject.h>
+#include <linux/memory_hotplug.h>
+#include <linux/mm.h>
+#include <linux/mmzone.h>
+#include <linux/pagemap.h>
+#include <linux/rmap.h>
+#include <linux/mmu_notifier.h>
+#include <linux/page_ext.h>
+#include <linux/page_idle.h>
+
+#include "internal.h"
+
+#define BITMAP_CHUNK_SIZE	sizeof(u64)
+#define BITMAP_CHUNK_BITS	(BITMAP_CHUNK_SIZE * BITS_PER_BYTE)
+
+/*
+ * Idle page tracking only considers user memory pages, for other types of
+ * pages the idle flag is always unset and an attempt to set it is silently
+ * ignored.
+ *
+ * We treat a page as a user memory page if it is on an LRU list, because it is
+ * always safe to pass such a page to rmap_walk(), which is essential for idle
+ * page tracking. With such an indicator of user pages we can skip isolated
+ * pages, but since there are not usually many of them, it will hardly affect
+ * the overall result.
+ *
+ * This function tries to get a user memory page by pfn as described above.
+ */
+static struct folio *page_idle_get_folio(unsigned long pfn)
+{
+	struct page *page = pfn_to_online_page(pfn);
+	struct folio *folio;
+
+	if (!page || PageTail(page))
+		return NULL;
+
+	folio = page_folio(page);
+	if (!folio_test_lru(folio) || !folio_try_get(folio))
+		return NULL;
+	if (unlikely(page_folio(page) != folio || !folio_test_lru(folio))) {
+		folio_put(folio);
+		folio = NULL;
+	}
+	return folio;
+}
+
+static bool page_idle_clear_pte_refs_one(struct folio *folio,
+					struct vm_area_struct *vma,
+					unsigned long addr, void *arg)
+{
+	DEFINE_FOLIO_VMA_WALK(pvmw, folio, vma, addr, 0);
+	bool referenced = false;
+
+	while (page_vma_mapped_walk(&pvmw)) {
+		addr = pvmw.address;
+		if (pvmw.pte) {
+			/*
+			 * For PTE-mapped THP, one sub page is referenced,
+			 * the whole THP is referenced.
+			 *
+			 * PFN swap PTEs, such as device-exclusive ones, that
+			 * actually map pages are "old" from a CPU perspective.
+			 * The MMU notifier takes care of any device aspects.
+			 */
+			if (likely(pte_present(ptep_get(pvmw.pte))))
+				referenced |= ptep_test_and_clear_young(vma, addr, pvmw.pte);
+			referenced |= mmu_notifier_clear_young(vma->vm_mm, addr, addr + PAGE_SIZE);
+		} else if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE)) {
+			pmd_t pmdval = pmdp_get(pvmw.pmd);
+
+			if (likely(pmd_present(pmdval)))
+				referenced |= pmdp_test_and_clear_young(vma, addr, pvmw.pmd);
+			referenced |= mmu_notifier_clear_young(vma->vm_mm, addr, addr + PMD_SIZE);
+		} else {
+			/* unexpected pmd-mapped page? */
+			WARN_ON_ONCE(1);
+		}
+	}
+
+	if (referenced) {
+		folio_clear_idle(folio);
+		/*
+		 * We cleared the referenced bit in a mapping to this page. To
+		 * avoid interference with page reclaim, mark it young so that
+		 * folio_referenced() will return > 0.
+		 */
+		folio_set_young(folio);
+	}
+	return true;
+}
+
+static void page_idle_clear_pte_refs(struct folio *folio)
+{
+	/*
+	 * Since rwc.try_lock is unused, rwc is effectively immutable, so we
+	 * can make it static to save some cycles and stack.
+	 */
+	static struct rmap_walk_control rwc = {
+		.rmap_one = page_idle_clear_pte_refs_one,
+		.anon_lock = folio_lock_anon_vma_read,
+	};
+
+	if (!folio_mapped(folio) || !folio_raw_mapping(folio))
+		return;
+
+	if (!folio_trylock(folio))
+		return;
+
+	rmap_walk(folio, &rwc);
+	folio_unlock(folio);
+}
+
+static ssize_t page_idle_bitmap_read(struct file *file, struct kobject *kobj,
+				     const struct bin_attribute *attr, char *buf,
+				     loff_t pos, size_t count)
+{
+	u64 *out = (u64 *)buf;
+	struct folio *folio;
+	unsigned long pfn, end_pfn;
+	int bit;
+
+	if (pos % BITMAP_CHUNK_SIZE || count % BITMAP_CHUNK_SIZE)
+		return -EINVAL;
+
+	pfn = pos * BITS_PER_BYTE;
+	if (pfn >= max_pfn)
+		return 0;
+
+	end_pfn = pfn + count * BITS_PER_BYTE;
+	if (end_pfn > max_pfn)
+		end_pfn = max_pfn;
+
+	for (; pfn < end_pfn; pfn++) {
+		bit = pfn % BITMAP_CHUNK_BITS;
+		if (!bit)
+			*out = 0ULL;
+		folio = page_idle_get_folio(pfn);
+		if (folio) {
+			if (folio_test_idle(folio)) {
+				/*
+				 * The page might have been referenced via a
+				 * pte, in which case it is not idle. Clear
+				 * refs and recheck.
+				 */
+				page_idle_clear_pte_refs(folio);
+				if (folio_test_idle(folio))
+					*out |= 1ULL << bit;
+			}
+			folio_put(folio);
+		}
+		if (bit == BITMAP_CHUNK_BITS - 1)
+			out++;
+		cond_resched();
+	}
+	return (char *)out - buf;
+}
+
+static ssize_t page_idle_bitmap_write(struct file *file, struct kobject *kobj,
+				      const struct bin_attribute *attr, char *buf,
+				      loff_t pos, size_t count)
+{
+	const u64 *in = (u64 *)buf;
+	struct folio *folio;
+	unsigned long pfn, end_pfn;
+	int bit;
+
+	if (pos % BITMAP_CHUNK_SIZE || count % BITMAP_CHUNK_SIZE)
+		return -EINVAL;
+
+	pfn = pos * BITS_PER_BYTE;
+	if (pfn >= max_pfn)
+		return -ENXIO;
+
+	end_pfn = pfn + count * BITS_PER_BYTE;
+	if (end_pfn > max_pfn)
+		end_pfn = max_pfn;
+
+	for (; pfn < end_pfn; pfn++) {
+		bit = pfn % BITMAP_CHUNK_BITS;
+		if ((*in >> bit) & 1) {
+			folio = page_idle_get_folio(pfn);
+			if (folio) {
+				page_idle_clear_pte_refs(folio);
+				folio_set_idle(folio);
+				folio_put(folio);
+			}
+		}
+		if (bit == BITMAP_CHUNK_BITS - 1)
+			in++;
+		cond_resched();
+	}
+	return (char *)in - buf;
+}
+
+static const struct bin_attribute page_idle_bitmap_attr =
+		__BIN_ATTR(bitmap, 0600,
+			   page_idle_bitmap_read, page_idle_bitmap_write, 0);
+
+static const struct bin_attribute *const page_idle_bin_attrs[] = {
+	&page_idle_bitmap_attr,
+	NULL,
+};
+
+static const struct attribute_group page_idle_attr_group = {
+	.bin_attrs = page_idle_bin_attrs,
+	.name = "page_idle",
+};
+
+static int __init page_idle_init(void)
+{
+	int err;
+
+	err = sysfs_create_group(mm_kobj, &page_idle_attr_group);
+	if (err) {
+		pr_err("page_idle: register sysfs failed\n");
+		return err;
+	}
+	return 0;
+}
+subsys_initcall(page_idle_init);