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, 145 insertions, 0 deletions

diff --git a/arch/mips/bcm63xx/cs.c b/arch/mips/bcm63xx/cs.c
new file mode 100644
index 000000000..29205badc
--- /dev/null
+++ b/arch/mips/bcm63xx/cs.c
@@ -0,0 +1,145 @@
+/*
+ * 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) 2008 Maxime Bizon <mbizon@freebox.fr>
+ */
+
+#include <linux/kernel.h>
+#include <linux/errno.h>
+#include <linux/export.h>
+#include <linux/spinlock.h>
+#include <linux/log2.h>
+#include <bcm63xx_cpu.h>
+#include <bcm63xx_io.h>
+#include <bcm63xx_regs.h>
+#include <bcm63xx_cs.h>
+
+static DEFINE_SPINLOCK(bcm63xx_cs_lock);
+
+/*
+ * check if given chip select exists
+ */
+static int is_valid_cs(unsigned int cs)
+{
+	if (cs > 6)
+		return 0;
+	return 1;
+}
+
+/*
+ * Configure chipselect base address and size (bytes).
+ * Size must be a power of two between 8k and 256M.
+ */
+int bcm63xx_set_cs_base(unsigned int cs, u32 base, unsigned int size)
+{
+	unsigned long flags;
+	u32 val;
+
+	if (!is_valid_cs(cs))
+		return -EINVAL;
+
+	/* sanity check on size */
+	if (size != roundup_pow_of_two(size))
+		return -EINVAL;
+
+	if (size < 8 * 1024 || size > 256 * 1024 * 1024)
+		return -EINVAL;
+
+	val = (base & MPI_CSBASE_BASE_MASK);
+	/* 8k => 0 - 256M => 15 */
+	val |= (ilog2(size) - ilog2(8 * 1024)) << MPI_CSBASE_SIZE_SHIFT;
+
+	spin_lock_irqsave(&bcm63xx_cs_lock, flags);
+	bcm_mpi_writel(val, MPI_CSBASE_REG(cs));
+	spin_unlock_irqrestore(&bcm63xx_cs_lock, flags);
+
+	return 0;
+}
+
+EXPORT_SYMBOL(bcm63xx_set_cs_base);
+
+/*
+ * configure chipselect timing (ns)
+ */
+int bcm63xx_set_cs_timing(unsigned int cs, unsigned int wait,
+			   unsigned int setup, unsigned int hold)
+{
+	unsigned long flags;
+	u32 val;
+
+	if (!is_valid_cs(cs))
+		return -EINVAL;
+
+	spin_lock_irqsave(&bcm63xx_cs_lock, flags);
+	val = bcm_mpi_readl(MPI_CSCTL_REG(cs));
+	val &= ~(MPI_CSCTL_WAIT_MASK);
+	val &= ~(MPI_CSCTL_SETUP_MASK);
+	val &= ~(MPI_CSCTL_HOLD_MASK);
+	val |= wait << MPI_CSCTL_WAIT_SHIFT;
+	val |= setup << MPI_CSCTL_SETUP_SHIFT;
+	val |= hold << MPI_CSCTL_HOLD_SHIFT;
+	bcm_mpi_writel(val, MPI_CSCTL_REG(cs));
+	spin_unlock_irqrestore(&bcm63xx_cs_lock, flags);
+
+	return 0;
+}
+
+EXPORT_SYMBOL(bcm63xx_set_cs_timing);
+
+/*
+ * configure other chipselect parameter (data bus size, ...)
+ */
+int bcm63xx_set_cs_param(unsigned int cs, u32 params)
+{
+	unsigned long flags;
+	u32 val;
+
+	if (!is_valid_cs(cs))
+		return -EINVAL;
+
+	/* none of this fields apply to pcmcia */
+	if (cs == MPI_CS_PCMCIA_COMMON ||
+	    cs == MPI_CS_PCMCIA_ATTR ||
+	    cs == MPI_CS_PCMCIA_IO)
+		return -EINVAL;
+
+	spin_lock_irqsave(&bcm63xx_cs_lock, flags);
+	val = bcm_mpi_readl(MPI_CSCTL_REG(cs));
+	val &= ~(MPI_CSCTL_DATA16_MASK);
+	val &= ~(MPI_CSCTL_SYNCMODE_MASK);
+	val &= ~(MPI_CSCTL_TSIZE_MASK);
+	val &= ~(MPI_CSCTL_ENDIANSWAP_MASK);
+	val |= params;
+	bcm_mpi_writel(val, MPI_CSCTL_REG(cs));
+	spin_unlock_irqrestore(&bcm63xx_cs_lock, flags);
+
+	return 0;
+}
+
+EXPORT_SYMBOL(bcm63xx_set_cs_param);
+
+/*
+ * set cs status (enable/disable)
+ */
+int bcm63xx_set_cs_status(unsigned int cs, int enable)
+{
+	unsigned long flags;
+	u32 val;
+
+	if (!is_valid_cs(cs))
+		return -EINVAL;
+
+	spin_lock_irqsave(&bcm63xx_cs_lock, flags);
+	val = bcm_mpi_readl(MPI_CSCTL_REG(cs));
+	if (enable)
+		val |= MPI_CSCTL_ENABLE_MASK;
+	else
+		val &= ~MPI_CSCTL_ENABLE_MASK;
+	bcm_mpi_writel(val, MPI_CSCTL_REG(cs));
+	spin_unlock_irqrestore(&bcm63xx_cs_lock, flags);
+	return 0;
+}
+
+EXPORT_SYMBOL(bcm63xx_set_cs_status);