[Devel,RHEL7,COMMIT] ms/sched: Fix race on toggling cfs_bandwidth_used

Submitted by Konstantin Khorenko on Sept. 8, 2016, 10:15 a.m.

Details

Message ID 201609081015.u88AFWU4003232@finist_cl7.x64_64.work.ct
State New
Series "sched: Fix race on toggling cfs_bandwidth_used"
Headers show

Commit Message

Konstantin Khorenko Sept. 8, 2016, 10:15 a.m.
The commit is pushed to "branch-rh7-3.10.0-327.28.2.vz7.17.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh7-3.10.0-327.28.2.vz7.17.4
------>
commit 624feddaed7790b486f5e2281b1ca05272e61a64
Author: Ben Segall <bsegall@google.com>
Date:   Thu Sep 8 14:15:32 2016 +0400

    ms/sched: Fix race on toggling cfs_bandwidth_used
    
    ms commit: 1ee14e6c8cddeeb8a490d7b54cd9016e4bb900b4
    
    When we transition cfs_bandwidth_used to false, any currently
    throttled groups will incorrectly return false from cfs_rq_throttled.
    While tg_set_cfs_bandwidth will unthrottle them eventually, currently
    running code (including at least dequeue_task_fair and
    distribute_cfs_runtime) will cause errors.
    
    Fix this by turning off cfs_bandwidth_used only after unthrottling all
    cfs_rqs.
    
    Tested: toggle bandwidth back and forth on a loaded cgroup. Caused
    crashes in minutes without the patch, hasn't crashed with it.
    
    Signed-off-by: Ben Segall <bsegall@google.com>
    Signed-off-by: Peter Zijlstra <peterz@infradead.org>
    Cc: pjt@google.com
    Link: http://lkml.kernel.org/r/20131016181611.22647.80365.stgit@sword-of-the-dawn.mtv.corp.google.com
    Signed-off-by: Ingo Molnar <mingo@kernel.org>
    
    https://jira.sw.ru/browse/PSBM-51056
    
    Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
    Acked-by: Kirill Tkhai <ktkhai@virtuozzo.com>
---
 kernel/sched/core.c  |  9 ++++++++-
 kernel/sched/fair.c  | 16 +++++++++-------
 kernel/sched/sched.h |  3 ++-
 3 files changed, 19 insertions(+), 9 deletions(-)

Patch hide | download patch | download mbox

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 657b8e4..ccc826a 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -8470,7 +8470,12 @@  static int __tg_set_cfs_bandwidth(struct task_group *tg, u64 period, u64 quota)
 
 	runtime_enabled = quota != RUNTIME_INF;
 	runtime_was_enabled = cfs_b->quota != RUNTIME_INF;
-	account_cfs_bandwidth_used(runtime_enabled, runtime_was_enabled);
+	/*
+	 * If we need to toggle cfs_bandwidth_used, off->on must occur
+	 * before making related changes, and on->off must occur afterwards
+	 */
+	if (runtime_enabled && !runtime_was_enabled)
+		cfs_bandwidth_usage_inc();
 	raw_spin_lock_irq(&cfs_b->lock);
 	cfs_b->period = ns_to_ktime(period);
 	cfs_b->quota = quota;
@@ -8498,6 +8503,8 @@  static int __tg_set_cfs_bandwidth(struct task_group *tg, u64 period, u64 quota)
 	}
 	if (runtime_enabled != runtime_was_enabled)
 		tg_limit_toggled(tg);
+	if (runtime_was_enabled && !runtime_enabled)
+		cfs_bandwidth_usage_dec();
 	return ret;
 }
 
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 6d03ebf..be0c3dd 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -422,13 +422,14 @@  static inline bool cfs_bandwidth_used(void)
 	return static_key_false(&__cfs_bandwidth_used);
 }
 
-void account_cfs_bandwidth_used(int enabled, int was_enabled)
+void cfs_bandwidth_usage_inc(void)
 {
-	/* only need to count groups transitioning between enabled/!enabled */
-	if (enabled && !was_enabled)
-		static_key_slow_inc(&__cfs_bandwidth_used);
-	else if (!enabled && was_enabled)
-		static_key_slow_dec(&__cfs_bandwidth_used);
+	static_key_slow_inc(&__cfs_bandwidth_used);
+}
+
+void cfs_bandwidth_usage_dec(void)
+{
+	static_key_slow_dec(&__cfs_bandwidth_used);
 }
 #else /* HAVE_JUMP_LABEL */
 static bool cfs_bandwidth_used(void)
@@ -436,7 +437,8 @@  static bool cfs_bandwidth_used(void)
 	return true;
 }
 
-void account_cfs_bandwidth_used(int enabled, int was_enabled) {}
+void cfs_bandwidth_usage_inc(void) {}
+void cfs_bandwidth_usage_dec(void) {}
 #endif /* HAVE_JUMP_LABEL */
 
 static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index d9fe825..894260d 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1444,7 +1444,8 @@  extern void print_rt_stats(struct seq_file *m, int cpu);
 extern void init_cfs_rq(struct cfs_rq *cfs_rq);
 extern void init_rt_rq(struct rt_rq *rt_rq, struct rq *rq);
 
-extern void account_cfs_bandwidth_used(int enabled, int was_enabled);
+extern void cfs_bandwidth_usage_inc(void);
+extern void cfs_bandwidth_usage_dec(void);
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
 static inline struct cfs_rq *task_cfs_rq(struct task_struct *p)