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/acpi/hed.c | 90 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 90 insertions(+) create mode 100644 drivers/acpi/hed.c (limited to 'drivers/acpi/hed.c') diff --git a/drivers/acpi/hed.c b/drivers/acpi/hed.c new file mode 100644 index 000000000..48562f53d --- /dev/null +++ b/drivers/acpi/hed.c @@ -0,0 +1,90 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * ACPI Hardware Error Device (PNP0C33) Driver + * + * Copyright (C) 2010, Intel Corp. + * Author: Huang Ying + * + * ACPI Hardware Error Device is used to report some hardware errors + * notified via SCI, mainly the corrected errors. + */ + +#include +#include +#include +#include +#include +#include + +static const struct acpi_device_id acpi_hed_ids[] = { + {"PNP0C33", 0}, + {"", 0}, +}; +MODULE_DEVICE_TABLE(acpi, acpi_hed_ids); + +static bool hed_present; + +static BLOCKING_NOTIFIER_HEAD(acpi_hed_notify_list); + +int register_acpi_hed_notifier(struct notifier_block *nb) +{ + return blocking_notifier_chain_register(&acpi_hed_notify_list, nb); +} +EXPORT_SYMBOL_GPL(register_acpi_hed_notifier); + +void unregister_acpi_hed_notifier(struct notifier_block *nb) +{ + blocking_notifier_chain_unregister(&acpi_hed_notify_list, nb); +} +EXPORT_SYMBOL_GPL(unregister_acpi_hed_notifier); + +/* + * SCI to report hardware error is forwarded to the listeners of HED, + * it is used by HEST Generic Hardware Error Source with notify type + * SCI. + */ +static void acpi_hed_notify(acpi_handle handle, u32 event, void *data) +{ + blocking_notifier_call_chain(&acpi_hed_notify_list, 0, NULL); +} + +static int acpi_hed_probe(struct platform_device *pdev) +{ + int err; + + /* Only one hardware error device */ + if (hed_present) + return -EINVAL; + + err = devm_acpi_install_notify_handler(&pdev->dev, ACPI_DEVICE_NOTIFY, + acpi_hed_notify, NULL); + if (err) + return err; + + hed_present = true; + return 0; +} + +static void acpi_hed_remove(struct platform_device *pdev) +{ + hed_present = false; +} + +static struct platform_driver acpi_hed_driver = { + .probe = acpi_hed_probe, + .remove = acpi_hed_remove, + .driver = { + .name = "acpi-hardware-error-device", + .acpi_match_table = acpi_hed_ids, + }, +}; + +static int __init acpi_hed_driver_init(void) +{ + return platform_driver_register(&acpi_hed_driver); +} +subsys_initcall(acpi_hed_driver_init); + +MODULE_AUTHOR("Huang Ying"); +MODULE_DESCRIPTION("ACPI Hardware Error Device Driver"); +MODULE_LICENSE("GPL"); -- cgit v1.3.1