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, 101 insertions, 0 deletions

diff --git a/include/asm-generic/fixmap.h b/include/asm-generic/fixmap.h
new file mode 100644
index 000000000..3ff832ebc
--- /dev/null
+++ b/include/asm-generic/fixmap.h
@@ -0,0 +1,101 @@
+/*
+ * fixmap.h: compile-time virtual memory allocation
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 1998 Ingo Molnar
+ *
+ * Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
+ * x86_32 and x86_64 integration by Gustavo F. Padovan, February 2009
+ * Break out common bits to asm-generic by Mark Salter, November 2013
+ */
+
+#ifndef __ASM_GENERIC_FIXMAP_H
+#define __ASM_GENERIC_FIXMAP_H
+
+#include <linux/bug.h>
+#include <linux/mm_types.h>
+
+#define __fix_to_virt(x)	(FIXADDR_TOP - ((x) << PAGE_SHIFT))
+#define __virt_to_fix(x)	((FIXADDR_TOP - ((x)&PAGE_MASK)) >> PAGE_SHIFT)
+
+#ifndef __ASSEMBLER__
+/*
+ * 'index to address' translation. If anyone tries to use the idx
+ * directly without translation, we catch the bug with a NULL-deference
+ * kernel oops. Illegal ranges of incoming indices are caught too.
+ */
+static __always_inline unsigned long fix_to_virt(const unsigned int idx)
+{
+	BUILD_BUG_ON(idx >= __end_of_fixed_addresses);
+	return __fix_to_virt(idx);
+}
+
+static inline unsigned long virt_to_fix(const unsigned long vaddr)
+{
+	BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
+	return __virt_to_fix(vaddr);
+}
+
+/*
+ * Provide some reasonable defaults for page flags.
+ * Not all architectures use all of these different types and some
+ * architectures use different names.
+ */
+#ifndef FIXMAP_PAGE_NORMAL
+#define FIXMAP_PAGE_NORMAL PAGE_KERNEL
+#endif
+#if !defined(FIXMAP_PAGE_RO) && defined(PAGE_KERNEL_RO)
+#define FIXMAP_PAGE_RO PAGE_KERNEL_RO
+#endif
+#ifndef FIXMAP_PAGE_NOCACHE
+#define FIXMAP_PAGE_NOCACHE PAGE_KERNEL_NOCACHE
+#endif
+#ifndef FIXMAP_PAGE_IO
+#define FIXMAP_PAGE_IO PAGE_KERNEL_IO
+#endif
+#ifndef FIXMAP_PAGE_CLEAR
+#define FIXMAP_PAGE_CLEAR __pgprot(0)
+#endif
+
+#ifndef set_fixmap
+#define set_fixmap(idx, phys)				\
+	__set_fixmap(idx, phys, FIXMAP_PAGE_NORMAL)
+#endif
+
+#ifndef clear_fixmap
+#define clear_fixmap(idx)			\
+	__set_fixmap(idx, 0, FIXMAP_PAGE_CLEAR)
+#endif
+
+/* Return a pointer with offset calculated */
+#define __set_fixmap_offset(idx, phys, flags)				\
+({									\
+	unsigned long ________addr;					\
+	__set_fixmap(idx, phys, flags);					\
+	________addr = fix_to_virt(idx) + ((phys) & (PAGE_SIZE - 1));	\
+	________addr;							\
+})
+
+#define set_fixmap_offset(idx, phys) \
+	__set_fixmap_offset(idx, phys, FIXMAP_PAGE_NORMAL)
+
+/*
+ * Some hardware wants to get fixmapped without caching.
+ */
+#define set_fixmap_nocache(idx, phys) \
+	__set_fixmap(idx, phys, FIXMAP_PAGE_NOCACHE)
+
+#define set_fixmap_offset_nocache(idx, phys) \
+	__set_fixmap_offset(idx, phys, FIXMAP_PAGE_NOCACHE)
+
+/*
+ * Some fixmaps are for IO
+ */
+#define set_fixmap_io(idx, phys) \
+	__set_fixmap(idx, phys, FIXMAP_PAGE_IO)
+
+#endif /* __ASSEMBLER__ */
+#endif /* __ASM_GENERIC_FIXMAP_H */