sched_setscheduler call fails with EPERM error

测试代码如下

#include <sched.h>
#include <stdio.h>
#include <errno.h>
void main()
{
    struct sched_param params;
    params.sched_priority = sched_get_priority_max(SCHED_FIFO);
    printf("max pri=%d\n\r",params.sched_priority);
    if (sched_setscheduler(getpid(), SCHED_FIFO, &params)) {
        printf("sched set pri error errno=%d\n\r",errno);
    }
return;
}

gcc编译生成可执行程序copy到qemu虚拟机上执行报错，即使是root用户执行也是报错，strace跟踪发现报错 errno=EPERM(1): Operation not permitted。调试查看内核代码(5.2版本)发现kernel/sched/core.c __sched_setscheduler函数的如下位置返回

#ifdef CONFIG_RT_GROUP_SCHED
        /*
         * Do not allow realtime tasks into groups that have no runtime
         * assigned.
         */
        if (rt_bandwidth_enabled() && rt_policy(policy) &&
                task_group(p)->rt_bandwidth.rt_runtime == 0 &&
                !task_group_is_autogroup(task_group(p))) {
            task_rq_unlock(rq, p, &rf);
              return -EPERM;
        }
#endif

task_group(p)->rt_bandwidth.rt_runtime == 0 条件为真，进一步分析，此处对应/sys/fs/cgroup/cpu/user.slice/cpu.rt_runtime_us，将cpu.rt_runtime_us修改为非0，不再报错。cgroup user.slice是由systemd创建的。

参考：
https://access.redhat.com/discussions/2950611
https://github.com/coreos/bugs/issues/410
https://juejin.im/post/6844903858116755463#heading-1
https://github.com/moby/moby/issues/13983