Commit 233e7f26 authored by Oleg Nesterov's avatar Oleg Nesterov Committed by Ingo Molnar

stop_machine: Ensure that a queued callback will be called before cpu_stop_park()

cpu_stop_queue_work() checks stopper->enabled before it queues the
work, but ->enabled == T can only guarantee cpu_stop_signal_done()
if we race with cpu_down().

This is not enough for stop_two_cpus() or stop_machine(), they will
deadlock if multi_cpu_stop() won't be called by one of the target
CPU's. stop_machine/stop_cpus are fine, they rely on stop_cpus_mutex.
But stop_two_cpus() has to check cpu_active() to avoid the same race
with hotplug, and this check is very unobvious and probably not even
correct if we race with cpu_up().

Change cpu_down() pass to clear ->enabled before cpu_stopper_thread()
flushes the pending ->works and returns with KTHREAD_SHOULD_PARK set.

Note also that smpboot_thread_call() calls cpu_stop_unpark() which
sets enabled == T at CPU_ONLINE stage, so this CPU can't go away until
cpu_stopper_thread() is called at least once. This all means that if
cpu_stop_queue_work() succeeds, we know that work->fn() will be called.
Signed-off-by: default avatarOleg Nesterov <oleg@redhat.com>
Signed-off-by: default avatarPeter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20151008145131.GA18139@redhat.comSigned-off-by: default avatarIngo Molnar <mingo@kernel.org>
parent 6af597de
......@@ -33,6 +33,7 @@ void stop_one_cpu_nowait(unsigned int cpu, cpu_stop_fn_t fn, void *arg,
struct cpu_stop_work *work_buf);
int stop_cpus(const struct cpumask *cpumask, cpu_stop_fn_t fn, void *arg);
int try_stop_cpus(const struct cpumask *cpumask, cpu_stop_fn_t fn, void *arg);
void stop_machine_park(int cpu);
#else /* CONFIG_SMP */
......
......@@ -344,7 +344,7 @@ static int take_cpu_down(void *_param)
/* Give up timekeeping duties */
tick_handover_do_timer();
/* Park the stopper thread */
kthread_park(current);
stop_machine_park((long)param->hcpu);
return 0;
}
......
......@@ -452,6 +452,18 @@ static void cpu_stopper_thread(unsigned int cpu)
}
}
void stop_machine_park(int cpu)
{
struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu);
/*
* Lockless. cpu_stopper_thread() will take stopper->lock and flush
* the pending works before it parks, until then it is fine to queue
* the new works.
*/
stopper->enabled = false;
kthread_park(stopper->thread);
}
extern void sched_set_stop_task(int cpu, struct task_struct *stop);
static void cpu_stop_create(unsigned int cpu)
......@@ -462,17 +474,8 @@ static void cpu_stop_create(unsigned int cpu)
static void cpu_stop_park(unsigned int cpu)
{
struct cpu_stopper *stopper = &per_cpu(cpu_stopper, cpu);
struct cpu_stop_work *work, *tmp;
unsigned long flags;
/* drain remaining works */
spin_lock_irqsave(&stopper->lock, flags);
list_for_each_entry_safe(work, tmp, &stopper->works, list) {
list_del_init(&work->list);
cpu_stop_signal_done(work->done, false);
}
stopper->enabled = false;
spin_unlock_irqrestore(&stopper->lock, flags);
WARN_ON(!list_empty(&stopper->works));
}
static void cpu_stop_unpark(unsigned int cpu)
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment