Locklessly doing is_idle_task(rq->curr) is only okay because of RCU protection. The older variant of the broken code checked rq->curr == rq->idle instead and therefore didn't need RCU.
Fixes: f6be8af1c95d sched: Add new API wake_up_if_idle() to wake up the idle cpu Cc: Chuansheng Liu <[email protected]> Signed-off-by: Andy Lutomirski <[email protected]> --- This fixes a 3.18 regression, I think. This code seems to only be called inside preempt_disable, so I think it's unlikely to be a problem in practice. Compile-tested only, although the kbuild bot is chewing on it. Changes from v1: - Cc the right people (whoops) - Add rcu_read_lock and rcu_read_unlock. kernel/sched/core.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 240157c13ddc..b7937512965f 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -1623,8 +1623,10 @@ void wake_up_if_idle(int cpu) struct rq *rq = cpu_rq(cpu); unsigned long flags; - if (!is_idle_task(rq->curr)) - return; + rcu_read_lock(); + + if (!is_idle_task(rcu_dereference(rq->curr))) + goto out; if (set_nr_if_polling(rq->idle)) { trace_sched_wake_idle_without_ipi(cpu); @@ -1635,6 +1637,9 @@ void wake_up_if_idle(int cpu) /* Else cpu is not in idle, do nothing here */ raw_spin_unlock_irqrestore(&rq->lock, flags); } + +out: + rcu_read_unlock(); } bool cpus_share_cache(int this_cpu, int that_cpu) -- 1.9.3 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
