Re: [PATCH 1/3] task: Add a count of task rcu users

From: Frederic Weisbecker <frederic@kernel.org>
To: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
	Oleg Nesterov <oleg@redhat.com>,
	Russell King - ARM Linux admin <linux@armlinux.org.uk>,
	Peter Zijlstra <peterz@infradead.org>,
	Chris Metcalf <cmetcalf@ezchip.com>,
	Christoph Lameter <cl@linux.com>, Kirill Tkhai <tkhai@yandex.ru>,
	Mike Galbraith <efault@gmx.de>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@kernel.org>,
	Linux List Kernel Mailing <linux-kernel@vger.kernel.org>,
	Davidlohr Bueso <dave@stgolabs.net>
Subject: Re: [PATCH 1/3] task: Add a count of task rcu users
Date: Wed, 4 Sep 2019 16:44:16 +0200	[thread overview]
Message-ID: <20190904144415.GB20391@lenoir> (raw)
In-Reply-To: <87ef0yt221.fsf_-_@x220.int.ebiederm.org>

On Mon, Sep 02, 2019 at 11:51:34PM -0500, Eric W. Biederman wrote:
> 
> Add a count of the number of rcu users (currently 1) of the task
> struct so that we can later add the scheduler case and get rid of the
> very subtle task_rcu_dereference, and just use rcu_dereference.
> 
> As suggested by Oleg have the count overlap rcu_head so that no
> additional space in task_struct is required.
> 
> Inspired-by: Linus Torvalds <torvalds@linux-foundation.org>
> Inspired-by: Oleg Nesterov <oleg@redhat.com>
> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
> ---
>  include/linux/sched.h      | 5 ++++-
>  include/linux/sched/task.h | 1 +
>  kernel/exit.c              | 7 ++++++-
>  kernel/fork.c              | 7 +++----
>  4 files changed, 14 insertions(+), 6 deletions(-)
> 
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index 9f51932bd543..99a4518b9b17 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1142,7 +1142,10 @@ struct task_struct {
>  
>  	struct tlbflush_unmap_batch	tlb_ubc;
>  
> -	struct rcu_head			rcu;
> +	union {
> +		refcount_t		rcu_users;
> +		struct rcu_head		rcu;

So what happens if, say:

           CPU 1                         CPU 2
   --------------------------------------------------------------
   rcu_read_lock()
   p = rcu_dereference(rq->task)
   if (refcount_inc_not_zero(p->rcu_users)) {
       .....
                                         release_task() {
                                             put_task_struct_rcu_user() {
                                                 call_rcu() {
                                                     queue rcu_head
                                                 }
                                             }
                                         }
       put_task_struct_rcu_user(); //here rcu_users has been overwritten

Thanks.