[PATCH] sched, autogroup: Fix failure when writing to cpu.rt_runtime_us

[Zefan Li]

This is how to reproduce the bug:

int main() {
    struct sched_param param = {.sched_priority=1};

    if (fork() > 0)
        exit(0);

    setsid();

    if (sched_setscheduler(0, SCHED_RR, &param) < 0){
        perror("failed to sched_setscheduler()");
        return -1;
    }

    while(1)
        ;
}

  # ./test
  # mount -t cgroup -o cpu xxx /cgroup
  # cat /cgroup/cpu.rt_runtime_us
  950000
  # echo 940000 > /cgroup/cpu.rt_runtime_us
  Device or Resource busy

An autogroup has been created and there's an RT task in it. RT tasks in
autogroups are redirected to the root group and task_group() should
return &root_task_group, but it's broken and returns the autogroup.

We should reset p->sched_task_group when changing a normal task to an
RT task which is in autogroup.

Fixes: 8323f26ce342 ("sched: Fix race in task_group()")
Cc: <stable@xxxxxxxxxxxxxxx> # 3.6+
Reported-by: Zhang Wei <zhangwei555@xxxxxxxxxx>
Signed-off-by: Zefan Li <lizefan@xxxxxxxxxx>
---
 kernel/sched/core.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 89e7283..fccde96 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3414,6 +3414,7 @@ static int __sched_setscheduler(struct task_struct *p,
 	const struct sched_class *prev_class;
 	struct rq *rq;
 	int reset_on_fork;
+	bool reset_task_group = false;
 
 	/* may grab non-irq protected spin_locks */
 	BUG_ON(in_interrupt());
@@ -3546,10 +3547,13 @@ change:
 		 * 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, &flags);
-			return -EPERM;
+		    task_group(p)->rt_bandwidth.rt_runtime == 0) {
+			if (!task_group_is_autogroup(task_group(p))) {
+				task_rq_unlock(rq, p, &flags);
+				return -EPERM;
+			} else {
+				reset_task_group = true;
+			}
 		}
 #endif
 #ifdef CONFIG_SMP
@@ -3615,6 +3619,9 @@ change:
 	prev_class = p->sched_class;
 	__setscheduler(rq, p, attr);
 
+	if (reset_task_group)
+		p->sched_task_group = &root_task_group;
+
 	if (running)
 		p->sched_class->set_curr_task(rq);
 	if (queued) {
-- 
1.8.0.2

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/