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, 315 insertions, 0 deletions

diff --git a/include/asm-generic/pgalloc.h b/include/asm-generic/pgalloc.h
new file mode 100644
index 000000000..051aa1331
--- /dev/null
+++ b/include/asm-generic/pgalloc.h
@@ -0,0 +1,315 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __ASM_GENERIC_PGALLOC_H
+#define __ASM_GENERIC_PGALLOC_H
+
+#ifdef CONFIG_MMU
+
+#define GFP_PGTABLE_KERNEL	(GFP_KERNEL | __GFP_ZERO | __GFP_SKIP_KASAN)
+#define GFP_PGTABLE_USER	(GFP_PGTABLE_KERNEL | __GFP_ACCOUNT)
+
+/**
+ * __pte_alloc_one_kernel - allocate memory for a PTE-level kernel page table
+ * @mm: the mm_struct of the current context
+ *
+ * This function is intended for architectures that need
+ * anything beyond simple page allocation.
+ *
+ * Return: pointer to the allocated memory or %NULL on error
+ */
+static inline pte_t *__pte_alloc_one_kernel_noprof(struct mm_struct *mm)
+{
+	struct ptdesc *ptdesc = pagetable_alloc_noprof(GFP_PGTABLE_KERNEL, 0);
+
+	if (!ptdesc)
+		return NULL;
+	if (!pagetable_pte_ctor(mm, ptdesc)) {
+		pagetable_free(ptdesc);
+		return NULL;
+	}
+
+	ptdesc_set_kernel(ptdesc);
+
+	return ptdesc_address(ptdesc);
+}
+#define __pte_alloc_one_kernel(...)	alloc_hooks(__pte_alloc_one_kernel_noprof(__VA_ARGS__))
+
+#ifndef __HAVE_ARCH_PTE_ALLOC_ONE_KERNEL
+/**
+ * pte_alloc_one_kernel - allocate memory for a PTE-level kernel page table
+ * @mm: the mm_struct of the current context
+ *
+ * Return: pointer to the allocated memory or %NULL on error
+ */
+static inline pte_t *pte_alloc_one_kernel_noprof(struct mm_struct *mm)
+{
+	return __pte_alloc_one_kernel_noprof(mm);
+}
+#define pte_alloc_one_kernel(...)	alloc_hooks(pte_alloc_one_kernel_noprof(__VA_ARGS__))
+#endif
+
+/**
+ * pte_free_kernel - free PTE-level kernel page table memory
+ * @mm: the mm_struct of the current context
+ * @pte: pointer to the memory containing the page table
+ */
+static inline void pte_free_kernel(struct mm_struct *mm, pte_t *pte)
+{
+	pagetable_dtor_free(virt_to_ptdesc(pte));
+}
+
+/**
+ * __pte_alloc_one - allocate memory for a PTE-level user page table
+ * @mm: the mm_struct of the current context
+ * @gfp: GFP flags to use for the allocation
+ *
+ * Allocate memory for a page table and ptdesc and runs pagetable_pte_ctor().
+ *
+ * This function is intended for architectures that need
+ * anything beyond simple page allocation or must have custom GFP flags.
+ *
+ * Return: `struct page` referencing the ptdesc or %NULL on error
+ */
+static inline pgtable_t __pte_alloc_one_noprof(struct mm_struct *mm, gfp_t gfp)
+{
+	struct ptdesc *ptdesc;
+
+	ptdesc = pagetable_alloc_noprof(gfp, 0);
+	if (!ptdesc)
+		return NULL;
+	if (!pagetable_pte_ctor(mm, ptdesc)) {
+		pagetable_free(ptdesc);
+		return NULL;
+	}
+
+	return ptdesc_page(ptdesc);
+}
+#define __pte_alloc_one(...)	alloc_hooks(__pte_alloc_one_noprof(__VA_ARGS__))
+
+#ifndef __HAVE_ARCH_PTE_ALLOC_ONE
+/**
+ * pte_alloc_one - allocate a page for PTE-level user page table
+ * @mm: the mm_struct of the current context
+ *
+ * Allocate memory for a page table and ptdesc and runs pagetable_pte_ctor().
+ *
+ * Return: `struct page` referencing the ptdesc or %NULL on error
+ */
+static inline pgtable_t pte_alloc_one_noprof(struct mm_struct *mm)
+{
+	return __pte_alloc_one_noprof(mm, GFP_PGTABLE_USER);
+}
+#define pte_alloc_one(...)	alloc_hooks(pte_alloc_one_noprof(__VA_ARGS__))
+#endif
+
+/*
+ * Should really implement gc for free page table pages. This could be
+ * done with a reference count in struct page.
+ */
+
+/**
+ * pte_free - free PTE-level user page table memory
+ * @mm: the mm_struct of the current context
+ * @pte_page: the `struct page` referencing the ptdesc
+ */
+static inline void pte_free(struct mm_struct *mm, struct page *pte_page)
+{
+	struct ptdesc *ptdesc = page_ptdesc(pte_page);
+
+	pagetable_dtor_free(ptdesc);
+}
+
+
+#if CONFIG_PGTABLE_LEVELS > 2
+
+#ifndef __HAVE_ARCH_PMD_ALLOC_ONE
+/**
+ * pmd_alloc_one - allocate memory for a PMD-level page table
+ * @mm: the mm_struct of the current context
+ *
+ * Allocate memory for a page table and ptdesc and runs pagetable_pmd_ctor().
+ *
+ * Allocations use %GFP_PGTABLE_USER in user context and
+ * %GFP_PGTABLE_KERNEL in kernel context.
+ *
+ * Return: pointer to the allocated memory or %NULL on error
+ */
+static inline pmd_t *pmd_alloc_one_noprof(struct mm_struct *mm, unsigned long addr)
+{
+	struct ptdesc *ptdesc;
+	gfp_t gfp = GFP_PGTABLE_USER;
+
+	if (mm == &init_mm)
+		gfp = GFP_PGTABLE_KERNEL;
+	ptdesc = pagetable_alloc_noprof(gfp, 0);
+	if (!ptdesc)
+		return NULL;
+	if (!pagetable_pmd_ctor(mm, ptdesc)) {
+		pagetable_free(ptdesc);
+		return NULL;
+	}
+
+	if (mm == &init_mm)
+		ptdesc_set_kernel(ptdesc);
+
+	return ptdesc_address(ptdesc);
+}
+#define pmd_alloc_one(...)	alloc_hooks(pmd_alloc_one_noprof(__VA_ARGS__))
+#endif
+
+#ifndef __HAVE_ARCH_PMD_FREE
+static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
+{
+	struct ptdesc *ptdesc = virt_to_ptdesc(pmd);
+
+	BUG_ON((unsigned long)pmd & (PAGE_SIZE-1));
+	pagetable_dtor_free(ptdesc);
+}
+#endif
+
+#endif /* CONFIG_PGTABLE_LEVELS > 2 */
+
+#if CONFIG_PGTABLE_LEVELS > 3
+
+static inline pud_t *__pud_alloc_one_noprof(struct mm_struct *mm, unsigned long addr)
+{
+	gfp_t gfp = GFP_PGTABLE_USER;
+	struct ptdesc *ptdesc;
+
+	if (mm == &init_mm)
+		gfp = GFP_PGTABLE_KERNEL;
+
+	ptdesc = pagetable_alloc_noprof(gfp, 0);
+	if (!ptdesc)
+		return NULL;
+
+	pagetable_pud_ctor(ptdesc);
+
+	if (mm == &init_mm)
+		ptdesc_set_kernel(ptdesc);
+
+	return ptdesc_address(ptdesc);
+}
+#define __pud_alloc_one(...)	alloc_hooks(__pud_alloc_one_noprof(__VA_ARGS__))
+
+#ifndef __HAVE_ARCH_PUD_ALLOC_ONE
+/**
+ * pud_alloc_one - allocate memory for a PUD-level page table
+ * @mm: the mm_struct of the current context
+ *
+ * Allocate memory for a page table using %GFP_PGTABLE_USER for user context
+ * and %GFP_PGTABLE_KERNEL for kernel context.
+ *
+ * Return: pointer to the allocated memory or %NULL on error
+ */
+static inline pud_t *pud_alloc_one_noprof(struct mm_struct *mm, unsigned long addr)
+{
+	return __pud_alloc_one_noprof(mm, addr);
+}
+#define pud_alloc_one(...)	alloc_hooks(pud_alloc_one_noprof(__VA_ARGS__))
+#endif
+
+static inline void __pud_free(struct mm_struct *mm, pud_t *pud)
+{
+	struct ptdesc *ptdesc = virt_to_ptdesc(pud);
+
+	BUG_ON((unsigned long)pud & (PAGE_SIZE-1));
+	pagetable_dtor_free(ptdesc);
+}
+
+#ifndef __HAVE_ARCH_PUD_FREE
+static inline void pud_free(struct mm_struct *mm, pud_t *pud)
+{
+	__pud_free(mm, pud);
+}
+#endif
+
+#endif /* CONFIG_PGTABLE_LEVELS > 3 */
+
+#if CONFIG_PGTABLE_LEVELS > 4
+
+static inline p4d_t *__p4d_alloc_one_noprof(struct mm_struct *mm, unsigned long addr)
+{
+	gfp_t gfp = GFP_PGTABLE_USER;
+	struct ptdesc *ptdesc;
+
+	if (mm == &init_mm)
+		gfp = GFP_PGTABLE_KERNEL;
+
+	ptdesc = pagetable_alloc_noprof(gfp, 0);
+	if (!ptdesc)
+		return NULL;
+
+	pagetable_p4d_ctor(ptdesc);
+
+	if (mm == &init_mm)
+		ptdesc_set_kernel(ptdesc);
+
+	return ptdesc_address(ptdesc);
+}
+#define __p4d_alloc_one(...)	alloc_hooks(__p4d_alloc_one_noprof(__VA_ARGS__))
+
+#ifndef __HAVE_ARCH_P4D_ALLOC_ONE
+static inline p4d_t *p4d_alloc_one_noprof(struct mm_struct *mm, unsigned long addr)
+{
+	return __p4d_alloc_one_noprof(mm, addr);
+}
+#define p4d_alloc_one(...)	alloc_hooks(p4d_alloc_one_noprof(__VA_ARGS__))
+#endif
+
+static inline void __p4d_free(struct mm_struct *mm, p4d_t *p4d)
+{
+	struct ptdesc *ptdesc = virt_to_ptdesc(p4d);
+
+	BUG_ON((unsigned long)p4d & (PAGE_SIZE-1));
+	pagetable_dtor_free(ptdesc);
+}
+
+#ifndef __HAVE_ARCH_P4D_FREE
+static inline void p4d_free(struct mm_struct *mm, p4d_t *p4d)
+{
+	if (!mm_p4d_folded(mm))
+		__p4d_free(mm, p4d);
+}
+#endif
+
+#endif /* CONFIG_PGTABLE_LEVELS > 4 */
+
+static inline pgd_t *__pgd_alloc_noprof(struct mm_struct *mm, unsigned int order)
+{
+	gfp_t gfp = GFP_PGTABLE_USER;
+	struct ptdesc *ptdesc;
+
+	if (mm == &init_mm)
+		gfp = GFP_PGTABLE_KERNEL;
+
+	ptdesc = pagetable_alloc_noprof(gfp, order);
+	if (!ptdesc)
+		return NULL;
+
+	pagetable_pgd_ctor(ptdesc);
+
+	if (mm == &init_mm)
+		ptdesc_set_kernel(ptdesc);
+
+	return ptdesc_address(ptdesc);
+}
+#define __pgd_alloc(...)	alloc_hooks(__pgd_alloc_noprof(__VA_ARGS__))
+
+static inline void __pgd_free(struct mm_struct *mm, pgd_t *pgd)
+{
+	struct ptdesc *ptdesc = virt_to_ptdesc(pgd);
+
+	BUG_ON((unsigned long)pgd & (PAGE_SIZE-1));
+	pagetable_dtor_free(ptdesc);
+}
+
+#ifndef __HAVE_ARCH_PGD_FREE
+static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
+{
+	__pgd_free(mm, pgd);
+}
+#endif
+
+#endif /* CONFIG_MMU */
+
+#endif /* __ASM_GENERIC_PGALLOC_H */