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, 122 insertions, 0 deletions

diff --git a/drivers/dma/dw/of.c b/drivers/dma/dw/of.c
new file mode 100644
index 000000000..523ca8068
--- /dev/null
+++ b/drivers/dma/dw/of.c
@@ -0,0 +1,122 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Platform driver for the Synopsys DesignWare DMA Controller
+ *
+ * Copyright (C) 2007-2008 Atmel Corporation
+ * Copyright (C) 2010-2011 ST Microelectronics
+ * Copyright (C) 2013 Intel Corporation
+ */
+
+#include <linux/of.h>
+#include <linux/of_dma.h>
+#include <linux/platform_device.h>
+
+#include "internal.h"
+
+static struct dma_chan *dw_dma_of_xlate(struct of_phandle_args *dma_spec,
+					struct of_dma *ofdma)
+{
+	struct dw_dma *dw = ofdma->of_dma_data;
+	struct dw_dma_slave slave = {
+		.dma_dev = dw->dma.dev,
+	};
+	dma_cap_mask_t cap;
+
+	if (dma_spec->args_count < 3 || dma_spec->args_count > 4)
+		return NULL;
+
+	slave.src_id = dma_spec->args[0];
+	slave.dst_id = dma_spec->args[0];
+	slave.m_master = dma_spec->args[1];
+	slave.p_master = dma_spec->args[2];
+	if (dma_spec->args_count >= 4)
+		slave.channels = dma_spec->args[3];
+
+	if (WARN_ON(slave.src_id >= DW_DMA_MAX_NR_REQUESTS ||
+		    slave.dst_id >= DW_DMA_MAX_NR_REQUESTS ||
+		    slave.m_master >= dw->pdata->nr_masters ||
+		    slave.p_master >= dw->pdata->nr_masters ||
+		    slave.channels >= BIT(dw->pdata->nr_channels)))
+		return NULL;
+
+	dma_cap_zero(cap);
+	dma_cap_set(DMA_SLAVE, cap);
+
+	/* TODO: there should be a simpler way to do this */
+	return dma_request_channel(cap, dw_dma_filter, &slave);
+}
+
+struct dw_dma_platform_data *dw_dma_parse_dt(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct dw_dma_platform_data *pdata;
+	u32 tmp, arr[DW_DMA_MAX_NR_MASTERS];
+	u32 nr_masters;
+	u32 nr_channels;
+
+	if (of_property_read_u32(np, "dma-masters", &nr_masters))
+		return NULL;
+	if (nr_masters < 1 || nr_masters > DW_DMA_MAX_NR_MASTERS)
+		return NULL;
+
+	if (of_property_read_u32(np, "dma-channels", &nr_channels))
+		return NULL;
+	if (nr_channels > DW_DMA_MAX_NR_CHANNELS)
+		return NULL;
+
+	pdata = devm_kzalloc(&pdev->dev, sizeof(*pdata), GFP_KERNEL);
+	if (!pdata)
+		return NULL;
+
+	pdata->nr_masters = nr_masters;
+	pdata->nr_channels = nr_channels;
+
+	of_property_read_u32(np, "chan_allocation_order", &pdata->chan_allocation_order);
+	of_property_read_u32(np, "chan_priority", &pdata->chan_priority);
+
+	of_property_read_u32(np, "block_size", &pdata->block_size);
+
+	/* Try deprecated property first */
+	if (!of_property_read_u32_array(np, "data_width", arr, nr_masters)) {
+		for (tmp = 0; tmp < nr_masters; tmp++)
+			pdata->data_width[tmp] = BIT(arr[tmp] & 0x07);
+	}
+
+	/* If "data_width" and "data-width" both provided use the latter one */
+	of_property_read_u32_array(np, "data-width", pdata->data_width, nr_masters);
+
+	memset32(pdata->multi_block, 1, nr_channels);
+	of_property_read_u32_array(np, "multi-block", pdata->multi_block, nr_channels);
+
+	memset32(pdata->max_burst, DW_DMA_MAX_BURST, nr_channels);
+	of_property_read_u32_array(np, "snps,max-burst-len", pdata->max_burst, nr_channels);
+
+	of_property_read_u32(np, "snps,dma-protection-control", &pdata->protctl);
+	if (pdata->protctl > CHAN_PROTCTL_MASK)
+		return NULL;
+
+	return pdata;
+}
+
+void dw_dma_of_controller_register(struct dw_dma *dw)
+{
+	struct device *dev = dw->dma.dev;
+	int ret;
+
+	if (!dev->of_node)
+		return;
+
+	ret = of_dma_controller_register(dev->of_node, dw_dma_of_xlate, dw);
+	if (ret)
+		dev_err(dev, "could not register of_dma_controller\n");
+}
+
+void dw_dma_of_controller_free(struct dw_dma *dw)
+{
+	struct device *dev = dw->dma.dev;
+
+	if (!dev->of_node)
+		return;
+
+	of_dma_controller_free(dev->of_node);
+}