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/md/dm-vdo/permassert.h | 45 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 45 insertions(+) create mode 100644 drivers/md/dm-vdo/permassert.h (limited to 'drivers/md/dm-vdo/permassert.h') diff --git a/drivers/md/dm-vdo/permassert.h b/drivers/md/dm-vdo/permassert.h new file mode 100644 index 000000000..c34f2ba65 --- /dev/null +++ b/drivers/md/dm-vdo/permassert.h @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Copyright 2023 Red Hat + */ + +#ifndef PERMASSERT_H +#define PERMASSERT_H + +#include + +#include "errors.h" + +/* Utilities for asserting that certain conditions are met */ + +#define STRINGIFY(X) #X + +/* + * A hack to apply the "warn if unused" attribute to an integral expression. + * + * Since GCC doesn't propagate the warn_unused_result attribute to conditional expressions + * incorporating calls to functions with that attribute, this function can be used to wrap such an + * expression. With optimization enabled, this function contributes no additional instructions, but + * the warn_unused_result attribute still applies to the code calling it. + */ +static inline int __must_check vdo_must_use(int value) +{ + return value; +} + +/* Assert that an expression is true and return an error if it is not. */ +#define VDO_ASSERT(expr, ...) vdo_must_use(__VDO_ASSERT(expr, __VA_ARGS__)) + +/* Log a message if the expression is not true. */ +#define VDO_ASSERT_LOG_ONLY(expr, ...) __VDO_ASSERT(expr, __VA_ARGS__) + +#define __VDO_ASSERT(expr, ...) \ + (likely(expr) ? VDO_SUCCESS \ + : vdo_assertion_failed(STRINGIFY(expr), __FILE__, __LINE__, __VA_ARGS__)) + +/* Log an assertion failure message. */ +int vdo_assertion_failed(const char *expression_string, const char *file_name, + int line_number, const char *format, ...) + __printf(4, 5); + +#endif /* PERMASSERT_H */ -- cgit v1.3.1