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, 188 insertions, 0 deletions

diff --git a/drivers/mfd/lpc_sch.c b/drivers/mfd/lpc_sch.c
new file mode 100644
index 000000000..d069165f9
--- /dev/null
+++ b/drivers/mfd/lpc_sch.c
@@ -0,0 +1,188 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ *  lpc_sch.c - LPC interface for Intel Poulsbo SCH
+ *
+ *  LPC bridge function of the Intel SCH contains many other
+ *  functional units, such as Interrupt controllers, Timers,
+ *  Power Management, System Management, GPIO, RTC, and LPC
+ *  Configuration Registers.
+ *
+ *  Copyright (c) 2010 CompuLab Ltd
+ *  Copyright (c) 2014 Intel Corp.
+ *  Author: Denis Turischev <denis@compulab.co.il>
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/errno.h>
+#include <linux/acpi.h>
+#include <linux/pci.h>
+#include <linux/mfd/core.h>
+
+#define SMBASE		0x40
+#define SMBUS_IO_SIZE	64
+
+#define GPIO_BASE	0x44
+#define GPIO_IO_SIZE	64
+#define GPIO_IO_SIZE_CENTERTON	128
+
+#define WDTBASE		0x84
+#define WDT_IO_SIZE	64
+
+enum sch_chipsets {
+	LPC_SCH = 0,		/* Intel Poulsbo SCH */
+	LPC_ITC,		/* Intel Tunnel Creek */
+	LPC_CENTERTON,		/* Intel Centerton */
+	LPC_QUARK_X1000,	/* Intel Quark X1000 */
+};
+
+struct lpc_sch_info {
+	unsigned int io_size_smbus;
+	unsigned int io_size_gpio;
+	unsigned int io_size_wdt;
+};
+
+static struct lpc_sch_info sch_chipset_info[] = {
+	[LPC_SCH] = {
+		.io_size_smbus = SMBUS_IO_SIZE,
+		.io_size_gpio = GPIO_IO_SIZE,
+	},
+	[LPC_ITC] = {
+		.io_size_smbus = SMBUS_IO_SIZE,
+		.io_size_gpio = GPIO_IO_SIZE,
+		.io_size_wdt = WDT_IO_SIZE,
+	},
+	[LPC_CENTERTON] = {
+		.io_size_smbus = SMBUS_IO_SIZE,
+		.io_size_gpio = GPIO_IO_SIZE_CENTERTON,
+		.io_size_wdt = WDT_IO_SIZE,
+	},
+	[LPC_QUARK_X1000] = {
+		.io_size_gpio = GPIO_IO_SIZE,
+		.io_size_wdt = WDT_IO_SIZE,
+	},
+};
+
+static const struct pci_device_id lpc_sch_ids[] = {
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_SCH_LPC), .driver_data = LPC_SCH },
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_ITC_LPC), .driver_data = LPC_ITC },
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_CENTERTON_ILB), .driver_data = LPC_CENTERTON },
+	{ PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_QUARK_X1000_ILB), .driver_data = LPC_QUARK_X1000 },
+	{ }
+};
+MODULE_DEVICE_TABLE(pci, lpc_sch_ids);
+
+#define LPC_NO_RESOURCE		1
+#define LPC_SKIP_RESOURCE	2
+
+static int lpc_sch_get_io(struct pci_dev *pdev, int where, const char *name,
+			  struct resource *res, int size)
+{
+	unsigned int base_addr_cfg;
+	unsigned short base_addr;
+
+	if (size == 0)
+		return LPC_NO_RESOURCE;
+
+	pci_read_config_dword(pdev, where, &base_addr_cfg);
+	base_addr = 0;
+	if (!(base_addr_cfg & (1 << 31)))
+		dev_warn(&pdev->dev, "Decode of the %s I/O range disabled\n",
+			 name);
+	else
+		base_addr = (unsigned short)base_addr_cfg;
+
+	if (base_addr == 0) {
+		dev_warn(&pdev->dev, "I/O space for %s uninitialized\n", name);
+		return LPC_SKIP_RESOURCE;
+	}
+
+	res->start = base_addr;
+	res->end = base_addr + size - 1;
+	res->flags = IORESOURCE_IO;
+
+	return 0;
+}
+
+static int lpc_sch_populate_cell(struct pci_dev *pdev, int where,
+				 const char *name, int size, int id,
+				 struct mfd_cell *cell)
+{
+	struct resource *res;
+	int ret;
+
+	res = devm_kzalloc(&pdev->dev, sizeof(*res), GFP_KERNEL);
+	if (!res)
+		return -ENOMEM;
+
+	ret = lpc_sch_get_io(pdev, where, name, res, size);
+	if (ret)
+		return ret;
+
+	memset(cell, 0, sizeof(*cell));
+
+	cell->name = name;
+	cell->resources = res;
+	cell->num_resources = 1;
+	cell->ignore_resource_conflicts = true;
+	cell->id = id;
+
+	return 0;
+}
+
+static int lpc_sch_probe(struct pci_dev *dev, const struct pci_device_id *id)
+{
+	struct mfd_cell lpc_sch_cells[3];
+	struct lpc_sch_info *info = &sch_chipset_info[id->driver_data];
+	unsigned int cells = 0;
+	int ret;
+
+	ret = lpc_sch_populate_cell(dev, SMBASE, "isch_smbus",
+				    info->io_size_smbus,
+				    id->device, &lpc_sch_cells[cells]);
+	if (ret < 0)
+		return ret;
+	if (ret == 0)
+		cells++;
+
+	ret = lpc_sch_populate_cell(dev, GPIO_BASE, "sch_gpio",
+				    info->io_size_gpio,
+				    id->device, &lpc_sch_cells[cells]);
+	if (ret < 0)
+		return ret;
+	if (ret == 0)
+		cells++;
+
+	ret = lpc_sch_populate_cell(dev, WDTBASE, "ie6xx_wdt",
+				    info->io_size_wdt,
+				    id->device, &lpc_sch_cells[cells]);
+	if (ret < 0)
+		return ret;
+	if (ret == 0)
+		cells++;
+
+	if (cells == 0) {
+		dev_err(&dev->dev, "All decode registers disabled.\n");
+		return -ENODEV;
+	}
+
+	return mfd_add_devices(&dev->dev, 0, lpc_sch_cells, cells, NULL, 0, NULL);
+}
+
+static void lpc_sch_remove(struct pci_dev *dev)
+{
+	mfd_remove_devices(&dev->dev);
+}
+
+static struct pci_driver lpc_sch_driver = {
+	.name		= "lpc_sch",
+	.id_table	= lpc_sch_ids,
+	.probe		= lpc_sch_probe,
+	.remove		= lpc_sch_remove,
+};
+
+module_pci_driver(lpc_sch_driver);
+
+MODULE_AUTHOR("Denis Turischev <denis@compulab.co.il>");
+MODULE_DESCRIPTION("LPC interface for Intel Poulsbo SCH");
+MODULE_LICENSE("GPL");