From 46a177fb01e52ec0e3f9eab9b217a0f7c8909eeb Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 9 Sep 2025 11:58:02 +0200 Subject: [PATCH] sched: Add locking comments to sched_class methods 'Document' the locking context the various sched_class methods are called under. Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Juri Lelli Acked-by: Tejun Heo Acked-by: Vincent Guittot --- kernel/sched/core.c | 6 +-- kernel/sched/sched.h | 108 ++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 105 insertions(+), 9 deletions(-) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index e932439ae6dab..8c55740b1e062 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -583,8 +583,8 @@ EXPORT_SYMBOL(__trace_set_current_state); * * p->on_rq <- { 0, 1 = TASK_ON_RQ_QUEUED, 2 = TASK_ON_RQ_MIGRATING }: * - * is set by activate_task() and cleared by deactivate_task(), under - * rq->lock. Non-zero indicates the task is runnable, the special + * is set by activate_task() and cleared by deactivate_task()/block_task(), + * under rq->lock. Non-zero indicates the task is runnable, the special * ON_RQ_MIGRATING state is used for migration without holding both * rq->locks. It indicates task_cpu() is not stable, see task_rq_lock(). * @@ -4162,7 +4162,7 @@ int try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) * __schedule(). See the comment for smp_mb__after_spinlock(). * * Form a control-dep-acquire with p->on_rq == 0 above, to ensure - * schedule()'s deactivate_task() has 'happened' and p will no longer + * schedule()'s block_task() has 'happened' and p will no longer * care about it's own p->state. See the comment in __schedule(). */ smp_acquire__after_ctrl_dep(); diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index ea2ea8fd65051..346214503452c 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -2345,8 +2345,7 @@ extern const u32 sched_prio_to_wmult[40]; /* * {de,en}queue flags: * - * DEQUEUE_SLEEP - task is no longer runnable - * ENQUEUE_WAKEUP - task just became runnable + * SLEEP/WAKEUP - task is no-longer/just-became runnable * * SAVE/RESTORE - an otherwise spurious dequeue/enqueue, done to ensure tasks * are in a known state which allows modification. Such pairs @@ -2359,11 +2358,18 @@ extern const u32 sched_prio_to_wmult[40]; * * MIGRATION - p->on_rq == TASK_ON_RQ_MIGRATING (used for DEADLINE) * + * DELAYED - de/re-queue a sched_delayed task + * + * CLASS - going to update p->sched_class; makes sched_change call the + * various switch methods. + * * ENQUEUE_HEAD - place at front of runqueue (tail if not specified) * ENQUEUE_REPLENISH - CBS (replenish runtime and postpone deadline) * ENQUEUE_MIGRATED - the task was migrated during wakeup * ENQUEUE_RQ_SELECTED - ->select_task_rq() was called * + * XXX SAVE/RESTORE in combination with CLASS doesn't really make sense, but + * SCHED_DEADLINE seems to rely on this for now. */ #define DEQUEUE_SLEEP 0x0001 /* Matches ENQUEUE_WAKEUP */ @@ -2409,14 +2415,50 @@ struct sched_class { int uclamp_enabled; #endif + /* + * move_queued_task/activate_task/enqueue_task: rq->lock + * ttwu_do_activate/activate_task/enqueue_task: rq->lock + * wake_up_new_task/activate_task/enqueue_task: task_rq_lock + * ttwu_runnable/enqueue_task: task_rq_lock + * proxy_task_current: rq->lock + * sched_change_end + */ void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags); + /* + * move_queued_task/deactivate_task/dequeue_task: rq->lock + * __schedule/block_task/dequeue_task: rq->lock + * proxy_task_current: rq->lock + * wait_task_inactive: task_rq_lock + * sched_change_begin + */ bool (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags); + + /* + * do_sched_yield: rq->lock + */ void (*yield_task) (struct rq *rq); + /* + * yield_to: rq->lock (double) + */ bool (*yield_to_task)(struct rq *rq, struct task_struct *p); + /* + * move_queued_task: rq->lock + * __migrate_swap_task: rq->lock + * ttwu_do_activate: rq->lock + * ttwu_runnable: task_rq_lock + * wake_up_new_task: task_rq_lock + */ void (*wakeup_preempt)(struct rq *rq, struct task_struct *p, int flags); + /* + * schedule/pick_next_task/prev_balance: rq->lock + */ int (*balance)(struct rq *rq, struct task_struct *prev, struct rq_flags *rf); + + /* + * schedule/pick_next_task: rq->lock + */ struct task_struct *(*pick_task)(struct rq *rq); /* * Optional! When implemented pick_next_task() should be equivalent to: @@ -2429,48 +2471,102 @@ struct sched_class { */ struct task_struct *(*pick_next_task)(struct rq *rq, struct task_struct *prev); + /* + * sched_change: + * __schedule: rq->lock + */ void (*put_prev_task)(struct rq *rq, struct task_struct *p, struct task_struct *next); void (*set_next_task)(struct rq *rq, struct task_struct *p, bool first); + /* + * select_task_rq: p->pi_lock + * sched_exec: p->pi_lock + */ int (*select_task_rq)(struct task_struct *p, int task_cpu, int flags); + /* + * set_task_cpu: p->pi_lock || rq->lock (ttwu like) + */ void (*migrate_task_rq)(struct task_struct *p, int new_cpu); + /* + * ttwu_do_activate: rq->lock + * wake_up_new_task: task_rq_lock + */ void (*task_woken)(struct rq *this_rq, struct task_struct *task); + /* + * do_set_cpus_allowed: task_rq_lock + sched_change + */ void (*set_cpus_allowed)(struct task_struct *p, struct affinity_context *ctx); + /* + * sched_set_rq_{on,off}line: rq->lock + */ void (*rq_online)(struct rq *rq); void (*rq_offline)(struct rq *rq); + /* + * push_cpu_stop: p->pi_lock && rq->lock + */ struct rq *(*find_lock_rq)(struct task_struct *p, struct rq *rq); + /* + * hrtick: rq->lock + * sched_tick: rq->lock + * sched_tick_remote: rq->lock + */ void (*task_tick)(struct rq *rq, struct task_struct *p, int queued); + /* + * sched_cgroup_fork: p->pi_lock + */ void (*task_fork)(struct task_struct *p); + /* + * finish_task_switch: no locks + */ void (*task_dead)(struct task_struct *p); + /* + * sched_change + */ void (*switching_from)(struct rq *this_rq, struct task_struct *task); void (*switched_from) (struct rq *this_rq, struct task_struct *task); void (*switching_to) (struct rq *this_rq, struct task_struct *task); void (*switched_to) (struct rq *this_rq, struct task_struct *task); - - void (*reweight_task)(struct rq *this_rq, struct task_struct *task, - const struct load_weight *lw); - u64 (*get_prio) (struct rq *this_rq, struct task_struct *task); void (*prio_changed) (struct rq *this_rq, struct task_struct *task, u64 oldprio); + /* + * set_load_weight: task_rq_lock + sched_change + * __setscheduler_parms: task_rq_lock + sched_change + */ + void (*reweight_task)(struct rq *this_rq, struct task_struct *task, + const struct load_weight *lw); + + /* + * sched_rr_get_interval: task_rq_lock + */ unsigned int (*get_rr_interval)(struct rq *rq, struct task_struct *task); + /* + * task_sched_runtime: task_rq_lock + */ void (*update_curr)(struct rq *rq); #ifdef CONFIG_FAIR_GROUP_SCHED + /* + * sched_change_group: task_rq_lock + sched_change + */ void (*task_change_group)(struct task_struct *p); #endif #ifdef CONFIG_SCHED_CORE + /* + * pick_next_task: rq->lock + * try_steal_cookie: rq->lock (double) + */ int (*task_is_throttled)(struct task_struct *p, int cpu); #endif }; -- 2.47.3