sched: Provide sched_set_fifo()

SCHED_FIFO (or any static priority scheduler) is a broken scheduler model; it is fundamentally incapable of resource management, the one thing an OS is actually supposed to do. It is impossible to compose static priority workloads. One cannot take two well designed and functional static priority workloads and mash them together and still expect them to work. Therefore it doesn't make sense to expose the priority field; the kernel is fundamentally incapable of setting a sensible value, it needs systems knowledge that it doesn't have. Take away sched_setschedule() / sched_setattr() from modules and replace them with: - sched_set_fifo(p); create a FIFO task (at prio 50) - sched_set_fifo_low(p); create a task higher than NORMAL, which ends up being a FIFO task at prio 1. - sched_set_normal(p, nice); (re)set the task to normal This stops the proliferation of randomly chosen, and irrelevant, FIFO priorities that dont't really mean anything anyway. The system administrator/integrator, whoever has insight into the actual system design and requirements (userspace) can set-up appropriate priorities if and when needed.

Peter Zijlstra committed 6 years ago

Commit 7318d4c1 parent 87e867b2 files · +50 −0Browse files

Unified · Split

2 files

include/linux/sched.h +3 −0
kernel/sched/core.c +47 −0

include/linux/sched.h+3−0

@@ -1653,6 +1653,9 @@ extern int idle_cpu(int cpu); · 1652 unmodified lines
1653	1653	extern int available_idle_cpu(int cpu);
1654	1654	extern int sched_setscheduler(struct task_struct , int, const struct sched_param );
1655	1655	extern int sched_setscheduler_nocheck(struct task_struct , int, const struct sched_param );
	1656	+extern int sched_set_fifo(struct task_struct *p);
	1657	+extern int sched_set_fifo_low(struct task_struct *p);
	1658	+extern int sched_set_normal(struct task_struct *p, int nice);
1656	1659	extern int sched_setattr(struct task_struct , const struct sched_attr );
1657	1660	extern int sched_setattr_nocheck(struct task_struct , const struct sched_attr );
1658	1661	extern struct task_struct *idle_task(int cpu);

kernel/sched/core.c+47−0

@@ -5124,6 +5124,8 @@ static int _sched_setscheduler(struct task_struct *p, int policy, · 5123 unmodified lines
5124	5124	* @policy: new policy.
5125	5125	* @param: structure containing the new RT priority.
5126	5126	*
	5127	+ * Use sched_set_fifo(), read its comment.
	5128	+ *
5127	5129	* Return: 0 on success. An error code otherwise.
5128	5130	*
5129	5131	* NOTE that the task may be already dead.
@@ -5166,6 +5168,51 @@ int sched_setscheduler_nocheck(struct task_struct *p, int policy, · 36 unmodified lines
5166	5168	}
5167	5169	EXPORT_SYMBOL_GPL(sched_setscheduler_nocheck);
5168	5170
	5171	+/*
	5172	+ * SCHED_FIFO is a broken scheduler model; that is, it is fundamentally
	5173	+ * incapable of resource management, which is the one thing an OS really should
	5174	+ * be doing.
	5175	+ *
	5176	+ * This is of course the reason it is limited to privileged users only.
	5177	+ *
	5178	+ * Worse still; it is fundamentally impossible to compose static priority
	5179	+ * workloads. You cannot take two correctly working static prio workloads
	5180	+ * and smash them together and still expect them to work.
	5181	+ *
	5182	+ * For this reason 'all' FIFO tasks the kernel creates are basically at:
	5183	+ *
	5184	+ * MAX_RT_PRIO / 2
	5185	+ *
	5186	+ * The administrator _MUST_ configure the system, the kernel simply doesn't
	5187	+ * know enough information to make a sensible choice.
	5188	+ */
	5189	+int sched_set_fifo(struct task_struct *p)
	5190	+{
	5191	+ struct sched_param sp = { .sched_priority = MAX_RT_PRIO / 2 };
	5192	+ return sched_setscheduler_nocheck(p, SCHED_FIFO, &sp);
	5193	+}
	5194	+EXPORT_SYMBOL_GPL(sched_set_fifo);
	5195	+
	5196	+/*
	5197	+ * For when you don't much care about FIFO, but want to be above SCHED_NORMAL.
	5198	+ */
	5199	+int sched_set_fifo_low(struct task_struct *p)
	5200	+{
	5201	+ struct sched_param sp = { .sched_priority = 1 };
	5202	+ return sched_setscheduler_nocheck(p, SCHED_FIFO, &sp);
	5203	+}
	5204	+EXPORT_SYMBOL_GPL(sched_set_fifo_low);
	5205	+
	5206	+int sched_set_normal(struct task_struct *p, int nice)
	5207	+{
	5208	+ struct sched_attr attr = {
	5209	+ .sched_policy = SCHED_NORMAL,
	5210	+ .sched_nice = nice,
	5211	+ };
	5212	+ return sched_setattr_nocheck(p, &attr);
	5213	+}
	5214	+EXPORT_SYMBOL_GPL(sched_set_normal);
	5215	+
5169	5216	static int
5170	5217	do_sched_setscheduler(pid_t pid, int policy, struct sched_param __user *param)
5171	5218	{