diff --git a/include/linux/sched.h b/include/linux/sched.h index dd1feff73c7d650a3832730fafbee8e1f5bab99d..6f8e0cb8b55b97db2ec4f4f90b174b3b024df191 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2173,6 +2173,12 @@ int sched_trace_rq_nr_running(struct rq *rq); const struct cpumask *sched_trace_rd_span(struct root_domain *rd); +#ifdef CONFIG_SCHED_CORE +extern void sched_core_free(struct task_struct *tsk); +#else +static inline void sched_core_free(struct task_struct *tsk) { } +#endif + #ifdef CONFIG_QOS_SCHED void sched_move_offline_task(struct task_struct *p); void sched_qos_offline_wait(void); diff --git a/kernel/fork.c b/kernel/fork.c index 0fb86b65ae60ca5e1fc9fdc757fa4151ec98d7e3..628c2b3fafedd61bab007ab4f5ecbb8cd37ef810 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -739,6 +739,7 @@ void __put_task_struct(struct task_struct *tsk) exit_creds(tsk); delayacct_tsk_free(tsk); put_signal_struct(tsk->signal); + sched_core_free(tsk); if (!profile_handoff_task(tsk)) free_task(tsk); diff --git a/kernel/sched/Makefile b/kernel/sched/Makefile index 5fc9c9b70862f72486b4b6e57711d530ceaeca18..978fcfca5871d76524efb5268288f59dd5e20f71 100644 --- a/kernel/sched/Makefile +++ b/kernel/sched/Makefile @@ -36,3 +36,4 @@ obj-$(CONFIG_CPU_FREQ_GOV_SCHEDUTIL) += cpufreq_schedutil.o obj-$(CONFIG_MEMBARRIER) += membarrier.o obj-$(CONFIG_CPU_ISOLATION) += isolation.o obj-$(CONFIG_PSI) += psi.o +obj-$(CONFIG_SCHED_CORE) += core_sched.o diff --git a/kernel/sched/core.c b/kernel/sched/core.c index d4598582240b3ef930891d1d0cd041e767bb6cd1..a485c8729f7caff883e6c3a871705d9450bca226 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -158,7 +158,7 @@ static inline int rb_sched_core_cmp(const void *key, const struct rb_node *node) return 0; } -static void sched_core_enqueue(struct rq *rq, struct task_struct *p) +void sched_core_enqueue(struct rq *rq, struct task_struct *p) { rq->core->core_task_seq++; @@ -168,14 +168,15 @@ static void sched_core_enqueue(struct rq *rq, struct task_struct *p) rb_add(&p->core_node, &rq->core_tree, rb_sched_core_less); } -static void sched_core_dequeue(struct rq *rq, struct task_struct *p) +void sched_core_dequeue(struct rq *rq, struct task_struct *p) { rq->core->core_task_seq++; - if (!p->core_cookie) + if (!sched_core_enqueued(p)) return; rb_erase(&p->core_node, &rq->core_tree); + RB_CLEAR_NODE(&p->core_node); } /* diff --git a/kernel/sched/core_sched.c b/kernel/sched/core_sched.c new file mode 100644 index 0000000000000000000000000000000000000000..8d0869a9eb8c3103a254b154652a1707a1de9db7 --- /dev/null +++ b/kernel/sched/core_sched.c @@ -0,0 +1,109 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include "sched.h" + +/* + * A simple wrapper around refcount. An allocated sched_core_cookie's + * address is used to compute the cookie of the task. + */ +struct sched_core_cookie { + refcount_t refcnt; +}; + +unsigned long sched_core_alloc_cookie(void) +{ + struct sched_core_cookie *ck = kmalloc(sizeof(*ck), GFP_KERNEL); + if (!ck) + return 0; + + refcount_set(&ck->refcnt, 1); + sched_core_get(); + + return (unsigned long)ck; +} + +void sched_core_put_cookie(unsigned long cookie) +{ + struct sched_core_cookie *ptr = (void *)cookie; + + if (ptr && refcount_dec_and_test(&ptr->refcnt)) { + kfree(ptr); + sched_core_put(); + } +} + +unsigned long sched_core_get_cookie(unsigned long cookie) +{ + struct sched_core_cookie *ptr = (void *)cookie; + + if (ptr) + refcount_inc(&ptr->refcnt); + + return cookie; +} + +/* + * sched_core_update_cookie - replace the cookie on a task + * @p: the task to update + * @cookie: the new cookie + * + * Effectively exchange the task cookie; caller is responsible for lifetimes on + * both ends. + * + * Returns: the old cookie + */ +unsigned long sched_core_update_cookie(struct task_struct *p, unsigned long cookie) +{ + unsigned long old_cookie; + struct rq_flags rf; + struct rq *rq; + bool enqueued; + + rq = task_rq_lock(p, &rf); + + /* + * Since creating a cookie implies sched_core_get(), and we cannot set + * a cookie until after we've created it, similarly, we cannot destroy + * a cookie until after we've removed it, we must have core scheduling + * enabled here. + */ + SCHED_WARN_ON((p->core_cookie || cookie) && !sched_core_enabled(rq)); + + enqueued = sched_core_enqueued(p); + if (enqueued) + sched_core_dequeue(rq, p); + + old_cookie = p->core_cookie; + p->core_cookie = cookie; + + if (enqueued) + sched_core_enqueue(rq, p); + + /* + * If task is currently running, it may not be compatible anymore after + * the cookie change, so enter the scheduler on its CPU to schedule it + * away. + */ + if (task_running(rq, p)) + resched_curr(rq); + + task_rq_unlock(rq, p, &rf); + + return old_cookie; +} + +static unsigned long sched_core_clone_cookie(struct task_struct *p) +{ + unsigned long cookie, flags; + + raw_spin_lock_irqsave(&p->pi_lock, flags); + cookie = sched_core_get_cookie(p->core_cookie); + raw_spin_unlock_irqrestore(&p->pi_lock, flags); + + return cookie; +} + +void sched_core_free(struct task_struct *p) +{ + sched_core_put_cookie(p->core_cookie); +} diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 05bcd51b25daafc3b1a594d02953ce010ce02bca..91f8795f0693d29a0f4de4ef2f248bf38c524dbd 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -1267,6 +1267,22 @@ static inline bool sched_group_cookie_match(struct rq *rq, extern void queue_core_balance(struct rq *rq); +static inline bool sched_core_enqueued(struct task_struct *p) +{ + return !RB_EMPTY_NODE(&p->core_node); +} + +extern void sched_core_enqueue(struct rq *rq, struct task_struct *p); +extern void sched_core_dequeue(struct rq *rq, struct task_struct *p); + +extern void sched_core_get(void); +extern void sched_core_put(void); + +extern unsigned long sched_core_alloc_cookie(void); +extern void sched_core_put_cookie(unsigned long cookie); +extern unsigned long sched_core_get_cookie(unsigned long cookie); +extern unsigned long sched_core_update_cookie(struct task_struct *p, unsigned long cookie); + #else /* !CONFIG_SCHED_CORE */ static inline bool sched_core_enabled(struct rq *rq)