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, 115 insertions, 0 deletions

diff --git a/arch/arm/lib/bitops.h b/arch/arm/lib/bitops.h
new file mode 100644
index 000000000..f069d1b23
--- /dev/null
+++ b/arch/arm/lib/bitops.h
@@ -0,0 +1,115 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include <asm/assembler.h>
+#include <asm/unwind.h>
+
+#if __LINUX_ARM_ARCH__ >= 6
+	.macro	bitop, name, instr
+ENTRY(	\name		)
+UNWIND(	.fnstart	)
+	ands	ip, r1, #3
+	strbne	r1, [ip]		@ assert word-aligned
+	mov	r2, #1
+	and	r3, r0, #31		@ Get bit offset
+	mov	r0, r0, lsr #5
+	add	r1, r1, r0, lsl #2	@ Get word offset
+#if __LINUX_ARM_ARCH__ >= 7 && defined(CONFIG_SMP)
+	.arch_extension	mp
+	ALT_SMP(W(pldw)	[r1])
+	ALT_UP(W(nop))
+#endif
+	mov	r3, r2, lsl r3
+1:	ldrex	r2, [r1]
+	\instr	r2, r2, r3
+	strex	r0, r2, [r1]
+	cmp	r0, #0
+	bne	1b
+	bx	lr
+UNWIND(	.fnend		)
+ENDPROC(\name		)
+	.endm
+
+	.macro	__testop, name, instr, store, barrier
+ENTRY(	\name		)
+UNWIND(	.fnstart	)
+	ands	ip, r1, #3
+	strbne	r1, [ip]		@ assert word-aligned
+	mov	r2, #1
+	and	r3, r0, #31		@ Get bit offset
+	mov	r0, r0, lsr #5
+	add	r1, r1, r0, lsl #2	@ Get word offset
+	mov	r3, r2, lsl r3		@ create mask
+	\barrier
+#if __LINUX_ARM_ARCH__ >= 7 && defined(CONFIG_SMP)
+	.arch_extension	mp
+	ALT_SMP(W(pldw)	[r1])
+	ALT_UP(W(nop))
+#endif
+1:	ldrex	r2, [r1]
+	ands	r0, r2, r3		@ save old value of bit
+	\instr	r2, r2, r3		@ toggle bit
+	strex	ip, r2, [r1]
+	cmp	ip, #0
+	bne	1b
+	\barrier
+	cmp	r0, #0
+	movne	r0, #1
+2:	bx	lr
+UNWIND(	.fnend		)
+ENDPROC(\name		)
+	.endm
+
+	.macro	testop, name, instr, store
+	__testop \name, \instr, \store, smp_dmb
+	.endm
+
+	.macro	sync_testop, name, instr, store
+	__testop \name, \instr, \store, __smp_dmb
+	.endm
+#else
+	.macro	bitop, name, instr
+ENTRY(	\name		)
+UNWIND(	.fnstart	)
+	ands	ip, r1, #3
+	strbne	r1, [ip]		@ assert word-aligned
+	and	r2, r0, #31
+	mov	r0, r0, lsr #5
+	mov	r3, #1
+	mov	r3, r3, lsl r2
+	save_and_disable_irqs ip
+	ldr	r2, [r1, r0, lsl #2]
+	\instr	r2, r2, r3
+	str	r2, [r1, r0, lsl #2]
+	restore_irqs ip
+	ret	lr
+UNWIND(	.fnend		)
+ENDPROC(\name		)
+	.endm
+
+/**
+ * testop - implement a test_and_xxx_bit operation.
+ * @instr: operational instruction
+ * @store: store instruction
+ *
+ * Note: we can trivially conditionalise the store instruction
+ * to avoid dirtying the data cache.
+ */
+	.macro	testop, name, instr, store
+ENTRY(	\name		)
+UNWIND(	.fnstart	)
+	ands	ip, r1, #3
+	strbne	r1, [ip]		@ assert word-aligned
+	and	r3, r0, #31
+	mov	r0, r0, lsr #5
+	save_and_disable_irqs ip
+	ldr	r2, [r1, r0, lsl #2]!
+	mov	r0, #1
+	tst	r2, r0, lsl r3
+	\instr	r2, r2, r0, lsl r3
+	\store	r2, [r1]
+	moveq	r0, #0
+	restore_irqs ip
+	ret	lr
+UNWIND(	.fnend		)
+ENDPROC(\name		)
+	.endm
+#endif