Skip to contentSkip to conversation

sched: Fix rt/dl hotplug condition

PATCH New Peter Zijlstra 2 participants 2 messages Feb 12, 2014 – Feb 21, 2014 1 message not in the archive Scheduler

lore
Share & tools
lore

Peter Zijlstra

author maintainer
#1 lore raw

PATCH 1/1 sched: Fix rt/dl hotplug condition

In reply to a message not in the archive

Dan Carpenter reported:

Quoted 2 lines

kernel/sched/rt.c:1347 pick_next_task_rt() warn: variable dereferenced before check 'prev' (see line 1338) kernel/sched/deadline.c:1011 pick_next_task_dl() warn: variable dereferenced before check 'prev' (see line 1005)

Cc Steven Rostedt <[email protected]>
Cc Juri Lelli <[email protected]>
Fixes 38033c37faab ("sched: Push down pre_schedule() and idle_balance()")
---
 kernel/sched/deadline.c | 2 +-
 kernel/sched/rt.c       | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
kernel/sched/deadline.c +1 -1
index ed31ef66ab9d..d60008e4bce9 100644
@@ -1001,7 +1001,7 @@ struct task_struct *pick_next_task_dl(struct rq *rq, struct task_struct *prev)
 	dl_rq = &rq->dl;
 
 #ifdef CONFIG_SMP
-	if (dl_task(prev))
+	if (prev && dl_task(prev))
 		pull_dl_task(rq);
 #endif
 
kernel/sched/rt.c +1 -1
index 72f9ec759972..7294b498b2c1 100644
@@ -1334,7 +1334,7 @@ pick_next_task_rt(struct rq *rq, struct task_struct *prev)
 
 #ifdef CONFIG_SMP
 	/* Try to pull RT tasks here if we lower this rq's prio */
-	if (rq->rt.highest_prio.curr > prev->prio)
+	if (prev && (rq->rt.highest_prio.curr > prev->prio))
 		pull_rt_task(rq);
 #endif
 
#2 lore raw

[tip:sched/core] sched: Fix hotplug task migration

sched: Fix hotplug task migration
Commit-ID:  3f1d2a318171bf61850d4e5a72031271e5aada76
Gitweb:     http://git.kernel.org/tip/3f1d2a318171bf61850d4e5a72031271e5aada76
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Wed, 12 Feb 2014 10:49:30 +0100
Committer:  Thomas Gleixner <[email protected]>
CommitDate: Fri, 21 Feb 2014 21:43:18 +0100

sched: Fix hotplug task migration

Dan Carpenter reported:

Quoted 2 lines

kernel/sched/rt.c:1347 pick_next_task_rt() warn: variable dereferenced before check 'prev' (see line 1338) kernel/sched/deadline.c:1011 pick_next_task_dl() warn: variable dereferenced before check 'prev' (see line 1005)

Kirill also spotted that migrate_tasks() will have an instant NULL deref because pick_next_task() will immediately deref prev.

Instead of fixing all the corner cases because migrate_tasks() can pass in a NULL prev task in the unlikely case of hot-un-plug, provide a fake task such that we can remove all the NULL checks from the far more common paths.

A further problem; not previously spotted; is that because we pushed pre_schedule() and idle_balance() into pick_next_task() we now need to avoid those getting called and pulling more tasks on our dying CPU.

We avoid pull_{dl,rt}_task() by setting fake_task.prio to MAX_PRIO+1. We also note that since we call pick_next_task() exactly the amount of times we have runnable tasks present, we should never land in idle_balance().

Fixes 38033c37faab ("sched: Push down pre_schedule() and idle_balance()")
Cc Juri Lelli <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Steven Rostedt <[email protected]>
---
 kernel/sched/core.c      | 18 +++++++++++++++++-
 kernel/sched/deadline.c  |  3 +--
 kernel/sched/fair.c      |  5 ++---
 kernel/sched/idle_task.c |  3 +--
 kernel/sched/rt.c        |  3 +--
 kernel/sched/sched.h     |  5 +++++
 kernel/sched/stop_task.c |  3 +--
 7 files changed, 28 insertions(+), 12 deletions(-)
kernel/sched/core.c +17 -1
index fb9764f..49db434 100644
@@ -4681,6 +4681,22 @@ static void calc_load_migrate(struct rq *rq)
 		atomic_long_add(delta, &calc_load_tasks);
 }
 
+static void put_prev_task_fake(struct rq *rq, struct task_struct *prev)
+{
+}
+
+static const struct sched_class fake_sched_class = {
+	.put_prev_task = put_prev_task_fake,
+};
+
+static struct task_struct fake_task = {
+	/*
+	 * Avoid pull_{rt,dl}_task()
+	 */
+	.prio = MAX_PRIO + 1,
+	.sched_class = &fake_sched_class,
+};
+
 /*
  * Migrate all tasks from the rq, sleeping tasks will be migrated by
  * try_to_wake_up()->select_task_rq().
@@ -4721,7 +4737,7 @@ static void migrate_tasks(unsigned int dead_cpu)
 		if (rq->nr_running == 1)
 			break;
 
-		next = pick_next_task(rq, NULL);
+		next = pick_next_task(rq, &fake_task);
 		BUG_ON(!next);
 		next->sched_class->put_prev_task(rq, next);
 
kernel/sched/deadline.c +1 -2
index ed31ef6..bfeb84e 100644
@@ -1008,8 +1008,7 @@ struct task_struct *pick_next_task_dl(struct rq *rq, struct task_struct *prev)
 	if (unlikely(!dl_rq->dl_nr_running))
 		return NULL;
 
-	if (prev)
-		prev->sched_class->put_prev_task(rq, prev);
+	put_prev_task(rq, prev);
 
 	dl_se = pick_next_dl_entity(rq, dl_rq);
 	BUG_ON(!dl_se);
kernel/sched/fair.c +2 -3
index 40c758b..e884e45 100644
@@ -4690,7 +4690,7 @@ again:
 	if (!cfs_rq->nr_running)
 		goto idle;
 
-	if (!prev || prev->sched_class != &fair_sched_class)
+	if (prev->sched_class != &fair_sched_class)
 		goto simple;
 
 	/*
@@ -4766,8 +4766,7 @@ simple:
 	if (!cfs_rq->nr_running)
 		goto idle;
 
-	if (prev)
-		prev->sched_class->put_prev_task(rq, prev);
+	put_prev_task(rq, prev);
 
 	do {
 		se = pick_next_entity(cfs_rq, NULL);
kernel/sched/idle_task.c +1 -2
index f7d03af..53ff9e7 100644
@@ -26,8 +26,7 @@ static void check_preempt_curr_idle(struct rq *rq, struct task_struct *p, int fl
 static struct task_struct *
 pick_next_task_idle(struct rq *rq, struct task_struct *prev)
 {
-	if (prev)
-		prev->sched_class->put_prev_task(rq, prev);
+	put_prev_task(rq, prev);
 
 	schedstat_inc(rq, sched_goidle);
 #ifdef CONFIG_SMP
kernel/sched/rt.c +1 -2
index 72f9ec7..65c2d68 100644
@@ -1344,8 +1344,7 @@ pick_next_task_rt(struct rq *rq, struct task_struct *prev)
 	if (rt_rq_throttled(rt_rq))
 		return NULL;
 
-	if (prev)
-		prev->sched_class->put_prev_task(rq, prev);
+	put_prev_task(rq, prev);
 
 	p = _pick_next_task_rt(rq);
 
kernel/sched/sched.h +5 -0
index 92018f9..d276147 100644
@@ -1147,6 +1147,11 @@ struct sched_class {
 #endif
 };
 
+static inline void put_prev_task(struct rq *rq, struct task_struct *prev)
+{
+	prev->sched_class->put_prev_task(rq, prev);
+}
+
 #define sched_class_highest (&stop_sched_class)
 #define for_each_class(class) \
    for (class = sched_class_highest; class; class = class->next)
kernel/sched/stop_task.c +1 -2
index a4147c9..d6ce65d 100644
@@ -31,8 +31,7 @@ pick_next_task_stop(struct rq *rq, struct task_struct *prev)
 	if (!stop || !stop->on_rq)
 		return NULL;
 
-	if (prev)
-		prev->sched_class->put_prev_task(rq, prev);
+	put_prev_task(rq, prev);
 
 	stop->se.exec_start = rq_clock_task(rq);