From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug 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 Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- drivers/reset/reset-scmi.c | 132 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 132 insertions(+) create mode 100644 drivers/reset/reset-scmi.c (limited to 'drivers/reset/reset-scmi.c') diff --git a/drivers/reset/reset-scmi.c b/drivers/reset/reset-scmi.c new file mode 100644 index 000000000..4335811e0 --- /dev/null +++ b/drivers/reset/reset-scmi.c @@ -0,0 +1,132 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * ARM System Control and Management Interface (ARM SCMI) reset driver + * + * Copyright (C) 2019-2021 ARM Ltd. + */ + +#include +#include +#include +#include +#include + +static const struct scmi_reset_proto_ops *reset_ops; + +/** + * struct scmi_reset_data - reset controller information structure + * @rcdev: reset controller entity + * @ph: ARM SCMI protocol handle used for communication with system controller + */ +struct scmi_reset_data { + struct reset_controller_dev rcdev; + const struct scmi_protocol_handle *ph; +}; + +#define to_scmi_reset_data(p) container_of((p), struct scmi_reset_data, rcdev) +#define to_scmi_handle(p) (to_scmi_reset_data(p)->ph) + +/** + * scmi_reset_assert() - assert device reset + * @rcdev: reset controller entity + * @id: ID of the reset to be asserted + * + * This function implements the reset driver op to assert a device's reset + * using the ARM SCMI protocol. + * + * Return: 0 for successful request, else a corresponding error value + */ +static int +scmi_reset_assert(struct reset_controller_dev *rcdev, unsigned long id) +{ + const struct scmi_protocol_handle *ph = to_scmi_handle(rcdev); + + return reset_ops->assert(ph, id); +} + +/** + * scmi_reset_deassert() - deassert device reset + * @rcdev: reset controller entity + * @id: ID of the reset to be deasserted + * + * This function implements the reset driver op to deassert a device's reset + * using the ARM SCMI protocol. + * + * Return: 0 for successful request, else a corresponding error value + */ +static int +scmi_reset_deassert(struct reset_controller_dev *rcdev, unsigned long id) +{ + const struct scmi_protocol_handle *ph = to_scmi_handle(rcdev); + + return reset_ops->deassert(ph, id); +} + +/** + * scmi_reset_reset() - reset the device + * @rcdev: reset controller entity + * @id: ID of the reset signal to be reset(assert + deassert) + * + * This function implements the reset driver op to trigger a device's + * reset signal using the ARM SCMI protocol. + * + * Return: 0 for successful request, else a corresponding error value + */ +static int +scmi_reset_reset(struct reset_controller_dev *rcdev, unsigned long id) +{ + const struct scmi_protocol_handle *ph = to_scmi_handle(rcdev); + + return reset_ops->reset(ph, id); +} + +static const struct reset_control_ops scmi_reset_ops = { + .assert = scmi_reset_assert, + .deassert = scmi_reset_deassert, + .reset = scmi_reset_reset, +}; + +static int scmi_reset_probe(struct scmi_device *sdev) +{ + struct scmi_reset_data *data; + struct device *dev = &sdev->dev; + struct device_node *np = dev->of_node; + const struct scmi_handle *handle = sdev->handle; + struct scmi_protocol_handle *ph; + + if (!handle) + return -ENODEV; + + reset_ops = handle->devm_protocol_get(sdev, SCMI_PROTOCOL_RESET, &ph); + if (IS_ERR(reset_ops)) + return PTR_ERR(reset_ops); + + data = devm_kzalloc(dev, sizeof(*data), GFP_KERNEL); + if (!data) + return -ENOMEM; + + data->rcdev.ops = &scmi_reset_ops; + data->rcdev.owner = THIS_MODULE; + data->rcdev.of_node = np; + data->rcdev.nr_resets = reset_ops->num_domains_get(ph); + data->ph = ph; + + return devm_reset_controller_register(dev, &data->rcdev); +} + +static const struct scmi_device_id scmi_id_table[] = { + { SCMI_PROTOCOL_RESET, "reset" }, + { }, +}; +MODULE_DEVICE_TABLE(scmi, scmi_id_table); + +static struct scmi_driver scmi_reset_driver = { + .name = "scmi-reset", + .probe = scmi_reset_probe, + .id_table = scmi_id_table, +}; +module_scmi_driver(scmi_reset_driver); + +MODULE_AUTHOR("Sudeep Holla "); +MODULE_DESCRIPTION("ARM SCMI reset controller driver"); +MODULE_LICENSE("GPL v2"); -- cgit v1.3.1