linux-rt-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] futex: Fix handling of bad requeue syscall pairing
@ 2009-08-07 21:45 Darren Hart
  2009-08-07 22:20 ` [PATCH V2] " Darren Hart
  0 siblings, 1 reply; 3+ messages in thread
From: Darren Hart @ 2009-08-07 21:45 UTC (permalink / raw)
  To: lkml, , linux-rt-users
  Cc: Thomas Gleixner, Peter Zijlstra, Steven Rostedt, Ingo Molnar,
	John Kacur, Eric Dumazet, Dinakar Guniguntala, John Stultz

From: Darren Hart <dvhltc@us.ibm.com>

If futex_requeue(requeue_pi=1) finds a futex_q that was created by a call
other than futex_wait_requeue_pi(), the q.rt_waiter will be null.  If so,
this will result in an oops from the following call graph:

futex_requeue()
  rt_mutex_start_proxy_lock()
    task_blocks_on_rt_mutex()
      waiter->task dereference
        OOPS

We currently WARN_ON() if this is detected, clearly this is inadequate.
If we detect a mis-pairing in futex_requeue(), bail out, sending -EINVAL
to user-space.

Signed-off-by: Darren Hart <dvhltc@us.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@elte.hu>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: John Kacur <jkacur@redhat.com>
CC: Dinakar Guniguntala <dino@in.ibm.com>
CC: John Stultz <johnstul@us.ibm.com>
---

 kernel/futex.c |   11 +++++++++--
 1 files changed, 9 insertions(+), 2 deletions(-)


diff --git a/kernel/futex.c b/kernel/futex.c
index 578c7b7..d806d6c 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -1306,8 +1306,15 @@ retry_private:
 		if (!match_futex(&this->key, &key1))
 			continue;
 
-		WARN_ON(!requeue_pi && this->rt_waiter);
-		WARN_ON(requeue_pi && !this->rt_waiter);
+		/*
+		 * FUTEX_WAIT_REQEUE_PI and FUTEX_CMP_REQUEUE_PI should always
+		 * be paired with each other and no other futex ops.
+		 */
+		if (requeue_pi && !this->rt_waiter ||
+		    !requeue_pi && this->rt_waiter) {
+			ret = -EINVAL;
+			break;
+		}
 
 		/*
 		 * Wake nr_wake waiters.  For requeue_pi, if we acquired the
-- 
Darren Hart
IBM Linux Technology Center
Real-Time Linux Team

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* [PATCH V2] futex: Fix handling of bad requeue syscall pairing
  2009-08-07 21:45 [PATCH] futex: Fix handling of bad requeue syscall pairing Darren Hart
@ 2009-08-07 22:20 ` Darren Hart
  2009-08-10 15:27   ` Peter Zijlstra
  0 siblings, 1 reply; 3+ messages in thread
From: Darren Hart @ 2009-08-07 22:20 UTC (permalink / raw)
  To: lkml, , linux-rt-users
  Cc: Thomas Gleixner, Peter Zijlstra, Steven Rostedt, Ingo Molnar,
	John Kacur, Eric Dumazet, Dinakar Guniguntala, John Stultz

From: Darren Hart <dvhltc@us.ibm.com>

If futex_requeue(requeue_pi=1) finds a futex_q that was created by a call
other the futex_wait_requeue_pi(), the q.rt_waiter may be null.  If so,
this will result in an oops from the following call graph:

futex_requeue()
  rt_mutex_start_proxy_lock()
    task_blocks_on_rt_mutex()
      waiter->task dereference
        OOPS

We currently WARN_ON() if this is detected, clearly this is inadequate.
If we detect a mispairing in futex_requeue(), bail out, seding -EINVAL to
user-space.

V2: Fix parenthesis warnings.

Signed-off-by: Darren Hart <dvhltc@us.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@elte.hu>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Dinakar Guniguntala <dino@in.ibm.com>
CC: John Stultz <johnstul@us.ibm.com>
---

 kernel/futex.c |   11 +++++++++--
 1 files changed, 9 insertions(+), 2 deletions(-)


diff --git a/kernel/futex.c b/kernel/futex.c
index df30983..4705d89 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -1306,8 +1306,15 @@ retry_private:
 		if (!match_futex(&this->key, &key1))
 			continue;
 
-		WARN_ON(!requeue_pi && this->rt_waiter);
-		WARN_ON(requeue_pi && !this->rt_waiter);
+		/*
+		 * FUTEX_WAIT_REQEUE_PI and FUTEX_CMP_REQUEUE_PI should always
+		 * be paired with each other and no other futex ops.
+		 */
+		if ((requeue_pi && !this->rt_waiter) ||
+		    (!requeue_pi && this->rt_waiter)) {
+			ret = -EINVAL;
+			break;
+		}
 
 		/*
 		 * Wake nr_wake waiters.  For requeue_pi, if we acquired the
-- 
Darren Hart
IBM Linux Technology Center
Real-Time Linux Team

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH V2] futex: Fix handling of bad requeue syscall pairing
  2009-08-07 22:20 ` [PATCH V2] " Darren Hart
@ 2009-08-10 15:27   ` Peter Zijlstra
  0 siblings, 0 replies; 3+ messages in thread
From: Peter Zijlstra @ 2009-08-10 15:27 UTC (permalink / raw)
  To: Darren Hart
  Cc: lkml,, linux-rt-users, Thomas Gleixner, Steven Rostedt,
	Ingo Molnar, John Kacur, Eric Dumazet, Dinakar Guniguntala,
	John Stultz

On Fri, 2009-08-07 at 15:20 -0700, Darren Hart wrote:
> From: Darren Hart <dvhltc@us.ibm.com>
> 
> If futex_requeue(requeue_pi=1) finds a futex_q that was created by a call
> other the futex_wait_requeue_pi(), the q.rt_waiter may be null.  If so,
> this will result in an oops from the following call graph:
> 
> futex_requeue()
>   rt_mutex_start_proxy_lock()
>     task_blocks_on_rt_mutex()
>       waiter->task dereference
>         OOPS
> 
> We currently WARN_ON() if this is detected, clearly this is inadequate.
> If we detect a mispairing in futex_requeue(), bail out, seding -EINVAL to
> user-space.
> 
> V2: Fix parenthesis warnings.
> 
> Signed-off-by: Darren Hart <dvhltc@us.ibm.com>
> Cc: Thomas Gleixner <tglx@linutronix.de>

Acked-by: Peter Zijlstra <peterz@infradead.org>

> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Ingo Molnar <mingo@elte.hu>
> CC: Eric Dumazet <eric.dumazet@gmail.com>
> CC: Dinakar Guniguntala <dino@in.ibm.com>
> CC: John Stultz <johnstul@us.ibm.com>
> ---
> 
>  kernel/futex.c |   11 +++++++++--
>  1 files changed, 9 insertions(+), 2 deletions(-)
> 
> 
> diff --git a/kernel/futex.c b/kernel/futex.c
> index df30983..4705d89 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -1306,8 +1306,15 @@ retry_private:
>  		if (!match_futex(&this->key, &key1))
>  			continue;
>  
> -		WARN_ON(!requeue_pi && this->rt_waiter);
> -		WARN_ON(requeue_pi && !this->rt_waiter);
> +		/*
> +		 * FUTEX_WAIT_REQEUE_PI and FUTEX_CMP_REQUEUE_PI should always
> +		 * be paired with each other and no other futex ops.
> +		 */
> +		if ((requeue_pi && !this->rt_waiter) ||
> +		    (!requeue_pi && this->rt_waiter)) {
> +			ret = -EINVAL;
> +			break;
> +		}
>  
>  		/*
>  		 * Wake nr_wake waiters.  For requeue_pi, if we acquired the

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2009-08-10 15:28 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-08-07 21:45 [PATCH] futex: Fix handling of bad requeue syscall pairing Darren Hart
2009-08-07 22:20 ` [PATCH V2] " Darren Hart
2009-08-10 15:27   ` Peter Zijlstra

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).