Re: [PATCH 10/16] rcu: Check for RCU_FLAG_GP_INIT bit in gp_flags for spurious wakeup

[Paul E. McKenney]

On Wed, Jul 23, 2014 at 01:09:47AM -0400, Pranith Kumar wrote:
> rcu_gp_init() is called once the gp_kthread wakes up on the condition that
> the init bit is set in the gp_flags. For a spurious wakeup we need to check that
> it is actually set. This commit ensures that the RCU_GP_FLAG_INIT bit is set in
> gp_flags.
> 
> Signed-off-by: Pranith Kumar <bobby.prani@xxxxxxxxx>

The trick here is that the other bits have no meaning if we are not yet
in a grace period.  If any get spuriously set, for example, by a race
between force_quiescent_state() and the end of a grace period, we really
do want to clear them all.

So sorry, but no.

							Thanx, Paul

> ---
>  kernel/rcu/tree.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
> index 8d598a2..f8847d9 100644
> --- a/kernel/rcu/tree.c
> +++ b/kernel/rcu/tree.c
> @@ -1600,7 +1600,7 @@ static int rcu_gp_init(struct rcu_state *rsp)
>  	rcu_bind_gp_kthread();
>  	raw_spin_lock_irq(&rnp->lock);
>  	smp_mb__after_unlock_lock();
> -	if (!ACCESS_ONCE(rsp->gp_flags)) {
> +	if (!(ACCESS_ONCE(rsp->gp_flags) & RCU_GP_FLAG_INIT)) {
>  		/* Spurious wakeup, tell caller to go back to sleep.  */
>  		raw_spin_unlock_irq(&rnp->lock);
>  		return 0;
> -- 
> 2.0.0.rc2
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/