Re: [RFC PATCH] rcu: using ACCESS_ONCE() to observe the jiffies_stall/rnp->qsmask value

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Tue, Sep 28, 2010 at 04:32:43PM +0800, Dongdong Deng wrote:
> Using ACCESS_ONCE() to observe the jiffies_stall/rnp->qsmask value
> due to the caller didn't hold the root_rcu/rnp node's lock.

Given that the quantities are only used once, ACCESS_ONCE() is strictly
speaking not necessary.  I am tempted to queue your patch regardless
for the documentation benefits.

Or is this patch necessary to fix a failure that you have been seeing?

In any case, thank you for taking a look at RCU!

							Thanx, Paul

> Signed-off-by: Dongdong Deng <dongdong.deng@windriver.com>
> CC: Dipankar Sarma <dipankar@in.ibm.com>
> CC: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> ---
>  kernel/rcutree.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/rcutree.c b/kernel/rcutree.c
> index d5bc439..07a7063 100644
> --- a/kernel/rcutree.c
> +++ b/kernel/rcutree.c
> @@ -532,9 +532,9 @@ static void check_cpu_stall(struct rcu_state *rsp, struct rcu_data *rdp)
> 
>  	if (rcu_cpu_stall_panicking)
>  		return;
> -	delta = jiffies - rsp->jiffies_stall;
> +	delta = jiffies - ACCESS_ONCE(rsp->jiffies_stall);
>  	rnp = rdp->mynode;
> -	if ((rnp->qsmask & rdp->grpmask) && delta >= 0) {
> +	if ((ACCESS_ONCE(rnp->qsmask) & rdp->grpmask) && delta >= 0) {
> 
>  		/* We haven't checked in, so go dump stack. */
>  		print_cpu_stall(rsp);
> -- 
> 1.6.0.4
>