Dl walleye 4.4 ab

include/linux/sched.h

@@ -1511,6 +1511,7 @@ struct sched_dl_entity {
 	 * code.
 	 */
 	int dl_throttled, dl_boosted, dl_yielded, dl_non_contending;
+	int dl_overrun;
 
 	/*
 	 * Bandwidth enforcement timer. Each -deadline task has its

include/uapi/linux/sched.h

@@ -49,6 +49,12 @@
  */
 #define SCHED_FLAG_RESET_ON_FORK	0x01
 #define SCHED_FLAG_RECLAIM		0x02
-#define SCHED_FLAG_SPECIAL		0x04
+#define SCHED_FLAG_DL_OVERRUN		0x04
+#define SCHED_FLAG_SPECIAL		0x08
+
+#define SCHED_FLAG_ALL (SCHED_FLAG_RESET_ON_FORK       | \
+                        SCHED_FLAG_RECLAIM             | \
+                        SCHED_FLAG_SPECIAL             | \
+                        SCHED_FLAG_DL_OVERRUN)
 
 #endif /* _UAPI_LINUX_SCHED_H */

kernel/sched/core.c

@@ -2233,6 +2233,7 @@ void __dl_clear_params(struct task_struct *p)
 	dl_se->dl_throttled = 0;
 	dl_se->dl_yielded = 0;
 	dl_se->dl_non_contending = 0;
+	dl_se->dl_overrun = 0;
 }
 
 /*
@@ -4006,14 +4007,29 @@ static void __setscheduler(struct rq *rq, struct task_struct *p,
 }
 
 static void
-__getparam_dl(struct task_struct *p, struct sched_attr *attr)
+__getparam_dl(struct task_struct *p, struct sched_attr *attr, unsigned int flags)
 {
 	struct sched_dl_entity *dl_se = &p->dl;
 
 	attr->sched_priority = p->rt_priority;
-	attr->sched_runtime = dl_se->dl_runtime;
-	attr->sched_deadline = dl_se->dl_deadline;
+
+	if (flags == 1 && p == current) {
+		update_curr_dl(task_rq(p));
+
+		/*
+		 * sched_runtime can never be negative because, since this
+		 * operation can be performed by the task on its own
+		 * sched_attr, if the bandwidth is <= 0, then the task is
+		 * throttled and therefore cannot perform the syscall.
+		 */
+		attr->sched_runtime = dl_se->runtime;
+		attr->sched_deadline = dl_se->deadline;
+	} else {
+		attr->sched_runtime = dl_se->dl_runtime;
+		attr->sched_deadline = dl_se->dl_deadline;
+	}
 	attr->sched_period = dl_se->dl_period;
+
 	attr->sched_flags = dl_se->flags;
 }
 
@@ -4119,10 +4135,7 @@ static int __sched_setscheduler(struct task_struct *p,
 			return -EINVAL;
 	}
 
-	if (attr->sched_flags &
-		~(SCHED_FLAG_RESET_ON_FORK |
-		  SCHED_FLAG_RECLAIM |
-		  SCHED_FLAG_SPECIAL))
+	if (attr->sched_flags & ~(SCHED_FLAG_ALL))
 		return -EINVAL;
 
 	/*
@@ -4281,7 +4294,7 @@ static int __sched_setscheduler(struct task_struct *p,
 	p->sched_reset_on_fork = reset_on_fork;
 	oldprio = p->prio;
 
-	if (pi) {
+	if (!dl_task(p) && pi) {
 		/*
 		 * Take priority boosted tasks into account. If the new
 		 * effective priority is unchanged, we just store the new
@@ -4683,7 +4696,7 @@ SYSCALL_DEFINE4(sched_getattr, pid_t, pid, struct sched_attr __user *, uattr,
 	int retval;
 
 	if (!uattr || pid < 0 || size > PAGE_SIZE ||
-	    size < SCHED_ATTR_SIZE_VER0 || flags)
+	    size < SCHED_ATTR_SIZE_VER0)
 		return -EINVAL;
 
 	rcu_read_lock();
@@ -4700,7 +4713,7 @@ SYSCALL_DEFINE4(sched_getattr, pid_t, pid, struct sched_attr __user *, uattr,
 	if (p->sched_reset_on_fork)
 		attr.sched_flags |= SCHED_FLAG_RESET_ON_FORK;
 	if (task_has_dl_policy(p))
-		__getparam_dl(p, &attr);
+		__getparam_dl(p, &attr, flags);
 	else if (task_has_rt_policy(p))
 		attr.sched_priority = p->rt_priority;
 	else

kernel/sched/cpufreq_schedutil.c

@@ -542,9 +542,13 @@ static int sugov_kthread_create(struct sugov_policy *sg_policy)
 		.sched_flags = SCHED_FLAG_SPECIAL,
 		.sched_nice = 0,
 		.sched_priority = 0,
-		.sched_runtime = 0,
-		.sched_deadline = 0,
-		.sched_period = 0,
+		/*
+		 * Fake (unused) bandwidth; workaround to "fix"
+		 * priority inheritance.
+		 */
+		.sched_runtime  =  1000000,
+		.sched_deadline = 10000000,
+		.sched_period   = 10000000,
 	};
 	struct cpufreq_policy *policy = sg_policy->policy;
 	int ret;

kernel/sched/deadline.c

@@ -48,6 +48,9 @@ void add_running_bw(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 {
 	u64 old = dl_rq->running_bw;
 
+	if (unlikely(dl_entity_is_special(dl_se)))
+		return;
+
 	lockdep_assert_held(&(rq_of_dl_rq(dl_rq))->lock);
 	dl_rq->running_bw += dl_se->dl_bw;
 	WARN_ON(dl_rq->running_bw < old); /* overflow */
@@ -61,6 +64,9 @@ void sub_running_bw(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 {
 	u64 old = dl_rq->running_bw;
 
+	if (unlikely(dl_entity_is_special(dl_se)))
+		return;
+
 	lockdep_assert_held(&(rq_of_dl_rq(dl_rq))->lock);
 	dl_rq->running_bw -= dl_se->dl_bw;
 	WARN_ON(dl_rq->running_bw > old); /* underflow */
@@ -75,6 +81,9 @@ void add_rq_bw(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 {
 	u64 old = dl_rq->this_bw;
 
+	if (unlikely(dl_entity_is_special(dl_se)))
+		return;
+
 	lockdep_assert_held(&(rq_of_dl_rq(dl_rq))->lock);
 	dl_rq->this_bw += dl_se->dl_bw;
 	WARN_ON(dl_rq->this_bw < old); /* overflow */
@@ -85,6 +94,9 @@ void sub_rq_bw(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 {
 	u64 old = dl_rq->this_bw;
 
+	if (unlikely(dl_entity_is_special(dl_se)))
+		return;
+
 	lockdep_assert_held(&(rq_of_dl_rq(dl_rq))->lock);
 	dl_rq->this_bw -= dl_se->dl_bw;
 	WARN_ON(dl_rq->this_bw > old); /* underflow */
@@ -866,7 +878,7 @@ u64 grub_reclaim(u64 delta, struct rq *rq, u64 u)
  * Update the current task's runtime statistics (provided it is still
  * a -deadline task and has not been removed from the dl_rq).
  */
-static void update_curr_dl(struct rq *rq)
+void update_curr_dl(struct rq *rq)
 {
 	struct task_struct *curr = rq->curr;
 	struct sched_dl_entity *dl_se = &curr->dl;
@@ -914,6 +926,12 @@ static void update_curr_dl(struct rq *rq)
 throttle:
 	if (dl_runtime_exceeded(dl_se) || dl_se->dl_yielded) {
 		dl_se->dl_throttled = 1;
+
+		/* If requested, inform the user about runtime overruns. */
+		if (dl_runtime_exceeded(dl_se) &&
+				(dl_se->flags & SCHED_FLAG_DL_OVERRUN))
+			dl_se->dl_overrun = 1;
+
 		__dequeue_task_dl(rq, curr, 0);
 		if (unlikely(dl_se->dl_boosted || !start_dl_timer(curr)))
 			enqueue_task_dl(rq, curr, ENQUEUE_REPLENISH);

kernel/sched/sched.h

@@ -84,6 +84,8 @@ static inline void update_cpu_load_active(struct rq *this_rq) { }
  */
 #define RUNTIME_INF	((u64)~0ULL)
 
+void update_curr_dl(struct rq *rq);
+
 static inline int idle_policy(int policy)
 {
 	return policy == SCHED_IDLE;
@@ -186,7 +188,7 @@ struct dl_bandwidth {
 
 static inline int dl_bandwidth_enabled(void)
 {
-	return sysctl_sched_rt_runtime >= 0;
+	return 0;
 }
 
 extern struct dl_bw *dl_bw_of(int i);

kernel/time/posix-cpu-timers.c

@@ -12,6 +12,7 @@
 #include <linux/random.h>
 #include <linux/tick.h>
 #include <linux/workqueue.h>
+#include <linux/sched/deadline.h>
 
 /*
  * Called after updating RLIMIT_CPU to run cpu timer and update
@@ -851,6 +852,14 @@ check_timers_list(struct list_head *timers,
 	return 0;
 }
 
+static inline void check_dl_overrun(struct task_struct *tsk)
+{
+	if (tsk->dl.dl_overrun) {
+		tsk->dl.dl_overrun = 0;
+		__group_send_sig_info(SIGXCPU, SEND_SIG_PRIV, tsk);
+	}
+}
+
 /*
  * Check for any per-thread CPU timers that have fired and move them off
  * the tsk->cpu_timers[N] list onto the firing list.  Here we update the
@@ -865,6 +874,9 @@ static void check_thread_timers(struct task_struct *tsk,
 	unsigned long long expires;
 	unsigned long soft;
 
+	if (dl_task(tsk))
+		check_dl_overrun(tsk);
+
 	/*
 	 * If cputime_expires is zero, then there are no active
 	 * per thread CPU timers.
@@ -969,6 +981,9 @@ static void check_process_timers(struct task_struct *tsk,
 	struct task_cputime cputime;
 	unsigned long soft;
 
+	if (dl_task(tsk))
+		check_dl_overrun(tsk);
+
 	/*
 	 * If cputimer is not running, then there are no active
 	 * process wide timers (POSIX 1.b, itimers, RLIMIT_CPU).
@@ -1175,6 +1190,9 @@ static inline int fastpath_timer_check(struct task_struct *tsk)
 			return 1;
 	}
 
+	if (dl_task(tsk) && tsk->dl.dl_overrun)
+		return 1;
+
 	return 0;
 }