Re: [PATCH] percpu_rwsem: fix missed wakeup due to reordering of load

From: Davidlohr Bueso
Date: Mon Dec 03 2018 - 01:38:45 EST


On 2018-11-30 07:10, Prateek Sood wrote:
In a scenario where cpu_hotplug_lock percpu_rw_semaphore is already
acquired for read operation by P1 using percpu_down_read().

Now we have P1 in the path of releaseing the cpu_hotplug_lock and P2
is in the process of acquiring cpu_hotplug_lock.

P1 P2
percpu_up_read() path percpu_down_write() path

rcu_sync_enter() //gp_state=GP_PASSED

rcu_sync_is_idle() //returns false down_write(rw_sem)

__percpu_up_read()

[L] task = rcu_dereference(w->task) //NULL

smp_rmb() [S] w->task = current

smp_mb()

[L] readers_active_check() //fails
schedule()

[S] __this_cpu_dec(read_count)

Since load of task can result in NULL. This can lead to missed wakeup
in rcuwait_wake_up(). Above sequence violated the following constraint
in rcuwait_wake_up():

WAIT WAKE
[S] tsk = current [S] cond = true
MB (A) MB (B)
[L] cond [L] tsk


Hmm yeah we don't want rcu_wake_up() to get hoisted over the __this_cpu_dec(read_count). The smp_rmb() does not make sense to me here in the first place. Did you run into this scenario by code inspection or you actually it the issue?

Thanks,
Davidlohr