Re: [PATCH 0/2] resource control file system - aka containers on top of nsproxy!

[Srivatsa Vaddagiri <vatsa@in.ibm.com>]

On Sat, Mar 03, 2007 at 02:21:00AM -0800, Paul Jackson wrote:
> Perhaps you could summarize what becomes of this hook, in this
> brave new world of rcfs ...

attach_task() still uses a synchronize_rcu before doing a put_nsproxy in
the rcfs patches. This means cpuset_update_task_memory_state() can read a 
task's cpuset->mems_generation under just a rcu_read_lock() (as it is
doing currently).

void cpuset_update_task_memory_state(void)
{
	tsk = current;
	...

        } else {
                rcu_read_lock();
                ns = rcu_dereference(tsk->nsproxy);
                my_cpusets_mem_gen = 
			ns->ctlr_data[cpuset_ctlr.subsys_id]->mems_generation;
                rcu_read_unlock();
        }

	...
}


-- 
Regards,
vatsa