linux-kernel - Re: [PATCH 1/8] rcu: add a warn to rcu_preempt_note_context

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Wed, 30 Oct 2013 04:02:44 -0700
From:	"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To:	Lai Jiangshan <laijs@...fujitsu.com>
Cc:	Steven Rostedt <rostedt@...dmis.org>,
	Peter Zijlstra <peterz@...radead.org>,
	linux-kernel@...r.kernel.org, Dipankar Sarma <dipankar@...ibm.com>
Subject: Re: [PATCH 1/8] rcu: add a warn to rcu_preempt_note_context_switch()

On Wed, Aug 07, 2013 at 06:24:57PM +0800, Lai Jiangshan wrote:
> It is expected that _nesting == INT_MIN if _nesting < 0.
> Add a warning to it if something unexpected happen.
> 
> Signed-off-by: Lai Jiangshan <laijs@...fujitsu.com>
> ---
>  kernel/rcutree_plugin.h |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
> 
> diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
> index 63098a5..8fd947e 100644
> --- a/kernel/rcutree_plugin.h
> +++ b/kernel/rcutree_plugin.h
> @@ -243,6 +243,7 @@ static void rcu_preempt_note_context_switch(int cpu)
>  				       : rnp->gpnum + 1);
>  		raw_spin_unlock_irqrestore(&rnp->lock, flags);
>  	} else if (t->rcu_read_lock_nesting < 0 &&
> +		   !WARN_ON_ONCE(t->rcu_read_lock_nesting != INT_MIN) &&

Finally getting back to this...

>From what I can see, this is safe right now because
->rcu_read_lock_nesting is incremented only in case of an interrupt, NMI,
or softirq interrupting the rcu_read_unlock() code path on the one hand,
and because the functions called from rcu_read_unlock_special() currently
disable interrupts before doing any rcu_read_lock()s.  With this in
mind, it is currently impossible to have a context switch occur within
an RCU read-side critical section that is invoked (either directly or
indirectly) from the portion of __rcu_read_unlock() that has negative
->rcu_read_lock_nesting.

But this could change should any part of the rt_mutex_unlock() code
called from rcu_read_unlock_special() were to do rcu_read_lock() before
disabling interrupts.  Is there any reason we should prohibit such a
pattern in rt_mutex_unlock()?

(For the record, I am currently OK prohibiting this pattern in
rcu_report_exp_rnp(), which is also called from rcu_read_unlock_special()
-- it seems unlikely that someone would use RCU to protect RCU's
expedited-grace-period data structures.)

							Thanx, Paul


>  		   t->rcu_read_unlock_special) {
> 
>  		/*
> -- 
> 1.7.4.4
> 

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