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, 159 insertions, 0 deletions

diff --git a/drivers/nvmem/mtk-efuse.c b/drivers/nvmem/mtk-efuse.c
new file mode 100644
index 000000000..00a84ea96
--- /dev/null
+++ b/drivers/nvmem/mtk-efuse.c
@@ -0,0 +1,159 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2015 MediaTek Inc.
+ * Author: Andrew-CT Chen <andrew-ct.chen@mediatek.com>
+ */
+
+#include <linux/device.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/nvmem-provider.h>
+#include <linux/platform_device.h>
+#include <linux/property.h>
+
+struct mtk_efuse_pdata {
+	bool uses_post_processing;
+};
+
+struct mtk_efuse_priv {
+	void __iomem *base;
+};
+
+static int mtk_reg_read(void *context,
+			unsigned int reg, void *_val, size_t bytes)
+{
+	struct mtk_efuse_priv *priv = context;
+	void __iomem *addr = priv->base + reg;
+	u8 *val = _val;
+	int i;
+
+	for (i = 0; i < bytes; i++, val++)
+		*val = readb(addr + i);
+
+	return 0;
+}
+
+static int mtk_efuse_gpu_speedbin_pp(void *context, const char *id, int index,
+				     unsigned int offset, void *data, size_t bytes)
+{
+	u8 *val = data;
+
+	if (val[0] < 8)
+		val[0] = BIT(val[0]);
+
+	return 0;
+}
+
+static void mtk_efuse_fixup_dt_cell_info(struct nvmem_device *nvmem,
+					 struct nvmem_cell_info *cell)
+{
+	size_t sz = strlen(cell->name);
+
+	/*
+	 * On some SoCs, the GPU speedbin is not read as bitmask but as
+	 * a number with range [0-7] (max 3 bits): post process to use
+	 * it in OPP tables to describe supported-hw.
+	 */
+	if (cell->nbits <= 3 &&
+	    strncmp(cell->name, "gpu-speedbin", min(sz, strlen("gpu-speedbin"))) == 0)
+		cell->read_post_process = mtk_efuse_gpu_speedbin_pp;
+}
+
+static int mtk_efuse_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct resource *res;
+	struct nvmem_device *nvmem;
+	struct nvmem_config econfig = {};
+	struct mtk_efuse_priv *priv;
+	const struct mtk_efuse_pdata *pdata;
+	struct platform_device *socinfo;
+
+	priv = devm_kzalloc(dev, sizeof(*priv), GFP_KERNEL);
+	if (!priv)
+		return -ENOMEM;
+
+	priv->base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
+	if (IS_ERR(priv->base))
+		return PTR_ERR(priv->base);
+
+	pdata = device_get_match_data(dev);
+	econfig.add_legacy_fixed_of_cells = true;
+	econfig.stride = 1;
+	econfig.word_size = 1;
+	econfig.reg_read = mtk_reg_read;
+	econfig.size = resource_size(res);
+	econfig.priv = priv;
+	econfig.dev = dev;
+	if (pdata->uses_post_processing)
+		econfig.fixup_dt_cell_info = &mtk_efuse_fixup_dt_cell_info;
+	nvmem = devm_nvmem_register(dev, &econfig);
+	if (IS_ERR(nvmem))
+		return PTR_ERR(nvmem);
+
+	socinfo = platform_device_register_data(&pdev->dev, "mtk-socinfo",
+						PLATFORM_DEVID_AUTO, NULL, 0);
+	if (IS_ERR(socinfo))
+		dev_info(dev, "MediaTek SoC Information will be unavailable\n");
+
+	platform_set_drvdata(pdev, socinfo);
+	return 0;
+}
+
+static const struct mtk_efuse_pdata mtk_mt8186_efuse_pdata = {
+	.uses_post_processing = true,
+};
+
+static const struct mtk_efuse_pdata mtk_efuse_pdata = {
+	.uses_post_processing = false,
+};
+
+static const struct of_device_id mtk_efuse_of_match[] = {
+	{ .compatible = "mediatek,mt8173-efuse", .data = &mtk_efuse_pdata },
+	{ .compatible = "mediatek,mt8186-efuse", .data = &mtk_mt8186_efuse_pdata },
+	{ .compatible = "mediatek,efuse", .data = &mtk_efuse_pdata },
+	{/* sentinel */},
+};
+MODULE_DEVICE_TABLE(of, mtk_efuse_of_match);
+
+static void mtk_efuse_remove(struct platform_device *pdev)
+{
+	struct platform_device *socinfo = platform_get_drvdata(pdev);
+
+	if (!IS_ERR_OR_NULL(socinfo))
+		platform_device_unregister(socinfo);
+}
+
+static struct platform_driver mtk_efuse_driver = {
+	.probe = mtk_efuse_probe,
+	.remove = mtk_efuse_remove,
+	.driver = {
+		.name = "mediatek,efuse",
+		.of_match_table = mtk_efuse_of_match,
+	},
+};
+
+static int __init mtk_efuse_init(void)
+{
+	int ret;
+
+	ret = platform_driver_register(&mtk_efuse_driver);
+	if (ret) {
+		pr_err("Failed to register efuse driver\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static void __exit mtk_efuse_exit(void)
+{
+	return platform_driver_unregister(&mtk_efuse_driver);
+}
+
+subsys_initcall(mtk_efuse_init);
+module_exit(mtk_efuse_exit);
+
+MODULE_AUTHOR("Andrew-CT Chen <andrew-ct.chen@mediatek.com>");
+MODULE_DESCRIPTION("Mediatek EFUSE driver");
+MODULE_LICENSE("GPL v2");