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/edac/debugfs.c | 91 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 91 insertions(+) create mode 100644 drivers/edac/debugfs.c (limited to 'drivers/edac/debugfs.c') diff --git a/drivers/edac/debugfs.c b/drivers/edac/debugfs.c new file mode 100644 index 000000000..447d0c620 --- /dev/null +++ b/drivers/edac/debugfs.c @@ -0,0 +1,91 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include "edac_module.h" + +static struct dentry *edac_debugfs; + +void __init edac_debugfs_init(void) +{ + edac_debugfs = debugfs_create_dir("edac", NULL); +} + +void edac_debugfs_exit(void) +{ + debugfs_remove_recursive(edac_debugfs); +} + +void edac_create_debugfs_nodes(struct mem_ctl_info *mci) +{ + mci->debugfs = debugfs_create_dir(mci->dev.kobj.name, edac_debugfs); +} + +/* Create a toplevel dir under EDAC's debugfs hierarchy */ +struct dentry *edac_debugfs_create_dir(const char *dirname) +{ + if (!edac_debugfs) + return NULL; + + return debugfs_create_dir(dirname, edac_debugfs); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_dir); + +/* Create a toplevel dir under EDAC's debugfs hierarchy with parent @parent */ +struct dentry * +edac_debugfs_create_dir_at(const char *dirname, struct dentry *parent) +{ + return debugfs_create_dir(dirname, parent); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_dir_at); + +/* + * Create a file under EDAC's hierarchy or a sub-hierarchy: + * + * @name: file name + * @mode: file permissions + * @parent: parent dentry. If NULL, it becomes the toplevel EDAC dir + * @data: private data of caller + * @fops: file operations of this file + */ +struct dentry * +edac_debugfs_create_file(const char *name, umode_t mode, struct dentry *parent, + void *data, const struct file_operations *fops) +{ + if (!parent) + parent = edac_debugfs; + + return debugfs_create_file(name, mode, parent, data, fops); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_file); + +/* Wrapper for debugfs_create_x8() */ +void edac_debugfs_create_x8(const char *name, umode_t mode, + struct dentry *parent, u8 *value) +{ + if (!parent) + parent = edac_debugfs; + + debugfs_create_x8(name, mode, parent, value); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_x8); + +/* Wrapper for debugfs_create_x16() */ +void edac_debugfs_create_x16(const char *name, umode_t mode, + struct dentry *parent, u16 *value) +{ + if (!parent) + parent = edac_debugfs; + + debugfs_create_x16(name, mode, parent, value); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_x16); + +/* Wrapper for debugfs_create_x32() */ +void edac_debugfs_create_x32(const char *name, umode_t mode, + struct dentry *parent, u32 *value) +{ + if (!parent) + parent = edac_debugfs; + + debugfs_create_x32(name, mode, parent, value); +} +EXPORT_SYMBOL_GPL(edac_debugfs_create_x32); -- cgit v1.3.1